Responses to Questions for the Record
U.S. House Committee on Energy and Commerce
Mr. Jack Dorsey, Chief Executive Officer, Twitter
The Honorable Greg Walden
1. Mr. Dorsey, you testified that Twitter uses thousands of "signals" to determine and
decide what to show, downrank, or filter. Please outline the specific behavioral signals that
were implicated in the search auto-suggest issue. Additionally, please outline the specific
signals that Twitter currently uses and how those signals determine what content to show,
downrank, or filter.
Twitter also uses a range of behavioral signals to determine how Tweets are organized
and presented in the home timeline, conversations, and search based on relevance. Twitter relies
on behavioral signals—such as how accounts behave and react to one another—to identify
content that detracts from a healthy public conversation, such as spam and abuse. Unless we
have determined that a Tweet violates Twitter policies, it will remain on the platform, and is
available in our product. Where we have identified a Tweet as potentially detracting from
healthy conversation (e.g.
, as potentially abusive), it will often only be available to view if you
click on “Show more replies” or choose to see everything in your search setting.
Some examples of behavioral signals we use, in combination with each other and a range
of other signals, to help identify this type of content include: an account with no confirmed email
address, simultaneous registration for multiple accounts, accounts that repeatedly Tweet and
mention accounts that do not follow them, or behavior that might indicate a coordinated attack.
Twitter is also examining how accounts are connected to those that violate our rules and how
they interact with each other. The accuracy and usefulness of the algorithms developed from
these behavioral signals will continue to improve over time.
These behavioral signals are an important factor in how Twitter organizes and presents
content in communal areas like conversation and search. Our primary goal is to ensure that
relevant content and Tweets contributing to healthy conversation will appear first in
conversations and search. Because our service operates in dozens of languages and hundreds of
cultural contexts around the globe, we have found that behavior is a strong signal that helps us
identify bad faith actors on our platform. The behavioral ranking that Twitter utilizes does not
consider in any way political views or ideology. It focuses solely on the behavior of all accounts.
Twitter is always working to improve our behavior-based ranking models such that their breadth
and accuracy will improve over time. We use thousands of behavioral signals in our
behavior-based ranking models—this ensures that no one signal drives the ranking outcomes and
protects against malicious attempts to manipulate our ranking systems.
2. Mr. Dorsey, you testified that “about 600,000 accounts” were impacted by the
autosuggest search issue. With respect to these accounts, please provide the following:
1
a. The number of Members of Congress' accounts that were impacted. Please
provide a complete list of all the names of Congressional members, or their
accounts, that were affected.
b. The number of Republican Members of Congress' accounts that were
impacted.
c. The number of Democratic Members of Congress' accounts that were
impacted.
d. The number of other Federal, State or local government accounts that were
impacted.
e. Please provide a complete list of all the names of the remaining
non-Congressional accounts.
In July 2018, we publicly acknowledged that some accounts were not being
auto-suggested even when people were searching for their specific name. Our technology was
using a decision-making criteria that considered the behavior of people following these accounts.
Specifically, the followers of these accounts had a higher proportion of abusive behavior on the
platform.
This issue impacted 600,000 accounts across the globe. The vast majority of impacted
accounts were not political in nature. The issue impacted 53 accounts of politicians in the U.S.,
representing 0.00883 percent of total affected accounts. This subset of affected accounts includes
10 official accounts of Republican Members of Congress. The remainder of impacted political
accounts relate to campaign activity and affected candidates across the political spectrum.
To be clear, this issue was limited solely to the search auto-suggestion function. The
accounts, their Tweets, and all surrounding conversation about those accounts remained
displayed in search results. Once identified, this issue was promptly resolved within 24 hours.
An analysis of accounts for Members of Congress that were affected by this search issue
demonstrates that there was no negative effect on the growth of their follower counts. To the
contrary, follower counts of impacted Members of Congress increased.
Twitter will not reveal the names of the impacted accounts due to privacy and security
concerns, but we can disclose information concerning the accounts of those who have publicly
discussed they were impacted.
We have enclosed information concerning follower count data displayed in graph format
for the official accounts of Congressmen Matt Gaetz, Jim Jordan, Mark Meadows, and Devin
Nunes. The graphs contain publicly available data on the follower counts for the three month
period surrounding the incident, which occurred in late July. The other impacted members of
Congress also saw their followers increase during this time period.
2
3. Mr. Dorsey, you testified that Twitter needs to do a better job with your Terms of
Service to make them more understandable and approachable. Please explain the steps
Twitter is taking or intends to take to rework the Twitter Terms of Service to accomplish
this?
Twitter’s Terms of Service (“Terms”) govern an individual’s access to and use of our
services, including our various websites, SMS, APIs, email notifications, applications, buttons,
widgets, ads, commerce services, and our other covered services. The Terms also cover any
information, text, links, graphics, photos, audio, videos, or other materials or arrangements of
materials uploaded, downloaded or appearing on the Services (collectively referred to as
“Content”).
By using the Services an individual who uses the platform agrees to be bound by these
Terms. Twitter may revise these Terms from time to time. The changes will not be retroactive,
and the most current version of the Terms, which will always be at twitter.com/tos, will govern
our relationship with the individual on the platform. We will try to notify individuals on our
platform of material revisions, for example via a service notification or an email to the email
associated with an account. By continuing to access or use the Services after those revisions
become effective, the individual agrees to be bound by the revised Terms.
Making a policy change requires in-depth research around trends in online behavior,
developing language that sets expectations around what’s allowed, and reviewer guidelines that
can be enforced across millions of Tweets. Once drafted, we gather feedback from our teams and
Trust & Safety Council. We gather input from around the world so that we can consider diverse,
global perspectives around the changing nature of online speech, including how our rules are
applied and interpreted in different cultural and social contexts. We then test the proposed rule
with samples of potentially abusive Tweets to measure the policy effectiveness and once we
determine it meets our expectations, build and operationalize product changes to support the
update. Finally, we train our global review teams, update the Twitter Rules, and start enforcing
it. In 2019, as in prior years, we intend to review the Twitter Terms of Service to determine how
to make the rules that govern our platform more approachable and accessible.
Twitter recently updated our Privacy Policy to include callouts, graphics, and animations
designed to enable people to better understand the data we receive, how it is used, and when it is
shared.
4. Mr. Dorsey, how does Twitter determine whether the information included in Twitter's
Terms of Service is understandable and approachable?
a. How does Twitter define understandable with respect to Twitter' s Terms of
Service?
b. How does twitter define approachable with respect to Twitter's Terms of
Services?
3
The Twitter Rules (along with all incorporated policies), Privacy Policy, and Terms of
Service collectively make up the "Twitter User Agreement" that governs an individual’s access
to and use of Twitter's services. We have the Twitter Rules in place to help ensure everyone feels
safe expressing their beliefs and we strive to enforce them with uniform consistency.
Our policies and enforcement options evolve continuously to address emerging behaviors
online and we sometimes come across instances where someone is reported for an incident that
took place prior to that behavior being prohibited. In those instances, we will generally require
the individual to delete the Tweet that violates the new rules but we won’t generally take other
enforcement action against them (e.g. suspension). This is reflective of the fact that the Twitter
Rules are a living document. We continue to expand and update both them and our enforcement
options to respond to the changing contours of online conversation. This is how we make Twitter
better for everyone.
We are continually working to update, refine, and improve both our enforcement and our
policies, informed by in-depth research around trends in online behavior both on and off Twitter,
feedback from the people who use Twitter, and input from a number of external entities,
including members of our Trust & Safety Council.
We believe we have to rely on a straight-forward, principled approach and focus on the
long term goal of understanding - not just in terms of the service itself - but in terms of the role
we play in society and our wider responsibility to foster and better serve a healthy public
conversation.
5. Mr. Dorsey, does Twitter ever review, use, or consider data or information about a user
unrelated to Twitter to make decisions about content posted by a Twitter user?
a. If yes, what data or information does Twitter consider and why?
Yes, but only in rare circumstances. A good example of this is our approach to violent
extremist groups. We take pride in Twitter being a platform where a diverse range of opinions
can be held and discussed, but we will not tolerate groups or individuals associated with them
who engage in and promote violence against civilians both on and off the platform. Accounts
affiliated with groups in which violence is a component of advancing their cause risk having a
chilling effect on opponents and bystanders. The violence that such groups promote could also
have dangerous consequences offline, jeopardizing their targets’ physical safety.
We prohibit the use of Twitter’s services by violent extremist groups – i.e., identified
groups subscribing to the use of violence as a means to advance their cause, whether political,
religious, or social. Groups that are prohibited on our services are those that identify as such or
engage in activity — both on and off the platform — that promotes violence. An individual on
Twitter may not affiliate with organizations that – whether by their own statements or activity
both on and off the platform – use or promote violence against civilians to further their causes.
4
We consider violent extremist groups to be those which identify through their stated
purpose, publications, or actions, as an extremist group; have engaged in, or currently engage in,
violence (and/or the promotion of violence) as a means to further their cause; and target civilians
in their acts (and/or promotion) of violence.
6. Mr. Dorsey, you testified that Twitter's verification program is not where you'd like it to
be and that it needs a "serious reboot." When was the verification program shut down?
Please explain the current problems and inadequacies with the verification program and
what steps Twitter is taking or intends to take to address this.
Verification was meant to authenticate identity and voice but it is interpreted as an
endorsement or an indicator of importance. We gave verified accounts visual prominence on the
service which deepened this perception. We should have addressed this earlier but did not
prioritize the work as we should have. This perception became worse when we opened up
verification for public submissions and verified people who we in no way endorse.
Twitter recognizes that we have created this confusion and need to resolve it. Beginning
in November 2017, we have paused all general verifications while we work to resolve this issue.
Twitter is currently working on a new authentication and verification program. In the
meantime, we are not accepting any public submissions for verification and have introduced new
guidelines for the program.
7. Mr. Dorsey, in Twitter’s July 26, 2018 blog post “Setting the record straight on shadow
banning,” company officials indicated:
“For the most part, we believe the issue had more to do with how other people were
interacting with these representatives' accounts than the accounts themselves
(see
bullet #3 above). There are communities that try to boost each other's presence on
the platform through coordinated engagement. We believe these types of actors
engaged with the representatives' accounts-- the impact of this coordinated behavior, in
combination with our implementation of search auto-suggestions, caused the
representatives' accounts to not show up in auto-suggestions.”
(emphasis added).
For additional context, the bullet no. 3 referenced above pointed out: “How other accounts
interact with you (e.g. who mutes you, who follows you, who retweets you, who blocks you,
etc)”.
Yet, in response to a question from Rep. Walberg about what “specific signals or actions of
other accounts interacting with the representatives’ account would you suggest contributed
to the auto-suggest issue,” you asserted that the “behaviors we were seeing were actual
violations of our terms of service.”
Please explain in detail whether the auto-suggest search issue was a consequence of
“interaction” issue described in Twitter's blog post or violations of Terms of Service as you
5
described in your testimony, or both? If it was a consequence of Terms of Service
violations, how and why would such violations render affected accounts invisible to Twitter
users?
In July 2018, we acknowledged that some accounts (including those of Republicans and
Democrats) were not being auto-suggested even when people were searching for their specific
name. Our usage of the behavioral signals within search was causing this to happen. Specifically,
if an account had a large number of followers who violated our terms of service, it impacted the
visibility of the account. To be clear, this only impacted our search auto-suggestions. The
accounts, their Tweets, and surrounding conversation about those accounts were still showing up
in search results. Once identified, this issue was promptly resolved within 24 hours. This
impacted 600,000 accounts across the globe and across the political spectrum. And most
accounts affected had nothing to do with politics at all. In addition to fixing the search
auto-suggestion function, Twitter is continuing to improve our systems so they can better detect
these issues and correct for them.
Twitter had made a change to how one of our behavior based algorithms works in search
results. When people used search, our algorithms were filtering out those that had a higher
likelihood of being abusive from the “Latest" tab by default. Those search results were visible in
“Latest” if someone turned off the quality filter in search, and they were also in Top search and
elsewhere throughout the product. Twitter decided that a higher level of precision is needed
when filtering to ensure these accounts are included in “Latest” by default. Twitter therefore
turned off the algorithm. As always, we will continue to refine our approach and will be
transparent about why we make the decisions that we do.
8. Mr. Dorsey, in response to a question about the Meghan McCain incident and the
inadequacies of Twitter's abuse prioritization mechanism, you indicated “[i]n this
particular case, the reason why was because [the violent and physical harm element] was
captured within an image rather than the tweet text itself”
(emphasis added). Is currently
Twitter without the technological tools to police harmful and abusive content embedded in
either images, .gifs, links, videos, and audio clips? If yes to any, how do human reviewers
police harmful and abusive content embedded in either images, .gifs, links, videos, and
audio clips?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
6
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
Twitter employs extensive content detection technology to identify and police harmful
and abusive content embedded in various forms of media on the platform. We use PhotoDNA
and hash matching technology, particularly in the context of child sexual exploitation material
and terrorism. We use these technologies to identify known violative content in order to surface
it for agent review, however, if it is the first time that an image has been seen, it would not
necessarily be subject to our technology. It is important to note that we continually expand our
databases of known violative content.
9. Mr. Dorsey, what policies and procedures does Twitter have in place to notify accounts
and users when their messages or content have been removed, suspended, banned, or
otherwise rendered invisible? What steps going forward does Twitter intend to take to
better notify accounts and users?
All individuals accessing or using Twitter’s services must adhere to the policies set forth
in the Twitter Rules. Failure to do so may result in Twitter taking one or more of the following
enforcement actions: (1) requiring an individual to delete prohibited content before he or she can
again create new posts and interact with other Twitter users; (2) temporarily limiting an
individual’s ability to create posts or interact with other Twitter users; (3) requesting an
individual verify account ownership with a phone number or email address; or (4) permanently
suspending an account or related accounts.
We have improved the transparency of our enforcement and believe that increasing
understanding of our rules can help educate people who use Twitter. For example one tool we
use to enforce our rules is putting people in a read-only state until they delete a Tweet that
violates our Terms of Service. Accounts that we put into period of limited functionality
subsequently generate 25% fewer abuse reports, and 65% of these accounts are in this read-only
state on only one instance.
Accounts under investigation or which have been detected as sharing content in violation
with the Twitter Rules may have their account or visibility limited in various parts of Twitter,
including search.
7
10. Mr. Dorsey, the Twitter Rules preclude users from posting graphic violence and adult
content on the Twitter platform.
a. Please provide Twitter’s definition of “graphic violence” and outline the
specific factors Twitter uses to determine whether content meets this
definition.
b. Please provide Twitter's definition of “adult content” and outline the specific
factors Twitter uses to determine whether content meets this definition.
Twitter allows some forms of graphic violence or adult content in Tweets marked as
containing sensitive media. We consider graphic violence to be any form of gory media related
to death, serious injury, violence, or surgical procedures. We consider adult content to be any
media that is pornographic and may be intended to cause sexual arousal.
Individuals on our platform may not use such content in live video, the person’s profile,
or header images. Additionally, Twitter may sometimes require an individual to remove
excessively graphic violence.
While we want people to feel empowered to share media that reflects their creativity or
individuality, or to show what’s happening in the world, we have heard feedback from people
that they don’t want to be exposed to sensitive media inadvertently. Additionally, research has
shown that repeated exposure to violent content online may negatively impact an individual’s
wellbeing.
For this reason, we place media containing adult content, graphic violence, or hateful
imagery behind an interstitial, or warning message. This enables us to advise viewers that they
may view sensitive media if they click through, helping those who do not wish to see it make an
informed decision. However, an individual can not include this type of content in live video, or
in profile or header images. All other instances of this sensitive content should be marked as
sensitive media.
11. Mr. Dorsey, how does Twitter review the platform to identify content that meets
Twitter’s definition of graphic violence or adult content?
a. Is content on the platform proactively scanned or reviewed by Twitter before
it is flagged by a Twitter user ?
i. If yes, is the screening done by algorithms or humans?
1. If the screening is done by algorithms, please outline the
factors and signals the algorithms use to determine what is
considered “graphic violence” and “adult content” as defined
by Twitter.
8
2. If the screening is done by humans, please provide the number
of employees responsible for reviewing content on the Twitter
platform to identify “graphic violence” or “adult content.”
ii. If no, why does Twitter solely rely on users to flag potentially violative
content?
We encourage individuals who Tweet media containing graphic violence and adult
content to mark their account as sharing potentially sensitive media via the safety page within
account settings. When this content is reported to our team for review, we will manually apply an
interstitial, or warning message. If an individual repeatedly uploads sensitive media that is
mislabeled, Twitter may permanently adjust that individual’s account settings to label media they
Tweet as containing material that may be sensitive.
Including sensitive media in live video, header or profile images is a violation of our
media policy. Those in violation will be required to remove this media from their account and
may be asked to verify their contact information or serve a period of time-out before they can use
their account again. Subsequent violations may result in permanent account suspension.
If an individual encounters media in Tweets that he or she believes should be treated as
sensitive under Twitter’s media policy, we ask the individual to report it by reporting a Tweet
and selecting that the Tweets displays a sensitive image. We subsequently provide
recommendations to the individual who reported the Tweet for additional actions he or she can
take to improve the Twitter experience. It is possible that if an individual sees something he or
she doesn’t like, and Twitter has not placed a warning label before it, it’s possible that it doesn’t
meet our threshold for a warning on the media.
Twitter reviews reports of media flagged by individuals using our platform to determine
if that media requires a warning message in order to comply with Twitter’s media policies. We
caution individuals that by flagging media they are bringing it to the attention of the Twitter
team. The flagged content will not automatically receive a warning message or be removed from
the site.
Twitter may also use automated techniques to detect and label potentially sensitive
media, and also utilize automated techniques to detect and label accounts that frequently tweet
potentially sensitive media.
12. Mr. Dorsey, the Twitter Rules indicate that some content that may be violative of the
Twitter Rules will be allowed but marked as “sensitive media.”
What factors does Twitter consider when deciding whether content should be
prohibited and removed from the Twitter platform or should be allowed and remain
on the platform, but marked as "sensitive media?"
Is this decision made by algorithms or humans?
9
i. If it is made by algorithms, what factors or signals do the algorithms
consider when making this decision?
ii. If it is made by humans, please provide the number of employees
responsible for making this decision and identify what factors those
employees consider.
Please see answers 10 and 11.
13. Mr. Dorsey, the Twitter Rules state that “at times, [Twitter] may prevent certain
content from trending” and that it may be content that either “violates Twitter Rules” or
“may attempt to manipulate trends.”
a. When content does not violate the Twitter Rules or attempt to manipulate
trends, does Twitter ever prevent such content from “trending?”
i. If so, please provide an example of content that Twitter has not
allowed to “trend” despite it being in compliance with the Twitter
Rules.
b. Please explain what factors Twitter considers when determining whether
content “attempts to manipulate trends.”
We want trends to promote healthy discussions on Twitter. This means that, at times, we
may prevent certain content from trending. These include trends that contain profanity or adult or
graphic references; incite hate on the basis of race, ethnicity, national origin, sexual orientation,
gender, gender identity, religious affiliation, age, disability, or disease; or violate the Twitter
Rules.
In certain cases, we may also consider the newsworthiness of the content, or if it is in the
public interest, when evaluating potential violations. In these cases, the content may continue to
trend on our platform. Note that even if we prevent the hashtag or the content from appearing on
the list of trending topics, an individual may still be able to access conversations around that
content on Twitter.
Keeping Twitter safe and free from malicious automation is a top priority for us. One of
the most common violations we see is the use of multiple accounts and the Twitter developer
platform to attempt to artificially amplify or inflate the prominence of certain Tweets. To be
clear: Twitter prohibits any attempt to use automation for the purposes of posting or
disseminating spam, and such behavior may result in enforcement actions.
In January 2018, we announced that as part of our Information Quality efforts we would
be making changes to TweetDeck and the Twitter Application Program Interface, or API, to limit
the ability of individuals to perform coordinated actions across multiple accounts. These changes
10
are an important step in ensuring we stay ahead of malicious activity targeting the crucial
conversations taking place on Twitter — including elections in the United States and around the
world.
Twitter does not allow developers and the individuals with whom they work to
simultaneously post identical or substantially similar content to multiple accounts. For example,
a service is not permitted to allow their users to select several accounts they control from which
to publish a given Tweet. This applies regardless of whether the Tweets are published to Twitter
at the same time, or are scheduled/queued for future publication. Bulk, aggressive, or very
high-volume automated Retweeting is not permitted under the Automation Rules, and may be
subject to enforcement actions. And simultaneously performing actions such as Likes, Retweets,
or follows from multiple accounts is also prohibited.
The use of any form of automation (including scheduling) to post identical or
substantially similar content, or to perform actions such as Likes or Retweets, across many
accounts is not permitted. For example, applications that coordinate activity across multiple
accounts to simultaneously post Tweets with a specific hashtag (e.g. in an attempt to cause that
topic to trend) are prohibited.
Posting duplicative or substantially similar content, replies, or mentions over multiple
accounts under a developer’s control, or creating duplicate or substantially similar accounts, with
or without the use of automation, is never allowed. Posting multiple updates (on a single account
or across multiple accounts a developer controls) to a trending or popular topic (for instance,
through the use of a specific hashtag) with an intent to subvert or manipulate the topic, or to
artificially inflate the prominence of a hashtag or topic, is never allowed.
14. Mr. Dorsey, the Twitter Rules provide that users may not engage in hateful conduct.
Please provide Twitter's definition of “hateful conduct” and outline the specific factors
Twitter uses to determine whether content meets this definition.
a. When content is flagged as “hateful conduct”, what factors does Twitter
consider when deciding whether to remove the content?
i. Does Twitter consider data points or information about a user
unrelated to the specific content to make this determination?
Twitter’s mission is to give everyone the power to create and share ideas and
information, and to express their opinions and beliefs without barriers. Free expression is a
human right – we believe that everyone has a voice, and the right to use it. Our role is to serve
the public conversation, which requires representation of a diverse range of perspectives.
We recognize that if people experience abuse on Twitter, it can jeopardize their ability to
express themselves. Research has shown that some groups of people are disproportionately
targeted with abuse online. This includes; women, people of color, lesbian, gay, bisexual,
transgender, queer, intersex, asexual individuals, marginalized and historically underrepresented
11
communities. For those who identity with multiple underrepresented groups, abuse may be more
common, more severe in nature and have a higher impact on those targeted.
We are committed to combating abuse motivated by hatred, prejudice or intolerance,
particularly abuse that seeks to silence the voices of those who have been historically
marginalized. For this reason, we prohibit behavior that targets individuals with abuse based on
protected category.
An individual on the platform is not permitted to promote violence against or directly
attack or threaten other people on the basis of race, ethnicity, national origin, sexual orientation,
gender, gender identity, religious affiliation, age, disability, or serious disease. We also do not
allow accounts whose primary purpose is inciting harm towards others on the basis of these
categories.
We do not allow individuals to use hateful images or symbols in their profile image or
profile header. Individuals on the platform as not allowed to use the username, display name, or
profile bio to engage in abusive behavior, such as targeted harassment or expressing hate towards
a person, group, or protected category.
Under this policy, we take action against behavior that targets individuals or an entire
protected category with hateful conduct. Targeting can happen in a number of ways, for
example, mentions, including a photo of an individual, referring to someone by their full name,
etc.
When determining the penalty for violating this policy, we consider a number of factors
including, but not limited to the severity of the violation and an individual’s previous record of
rule violations. For example, we may ask someone to remove the violating content and serve a
period of time in read-only mode before they can Tweet again. Subsequent violations will lead to
longer read-only periods and may eventually result in permanent account suspension. If an
account is engaging primarily in abusive behavior, or is deemed to have shared a violent threat,
we will permanently suspend the account upon initial review.
Please see the response to question five for additional information regarding our violent
extremist group policy.
15. Mr. Dorsey, when a Twitter user flags content as potentially violative of the Twitter
Rules, what specific steps does Twitter undertake to determine whether that content should
be removed from the platform?
In order to maintain a safe environment for individuals on Twitter, we may suspend
accounts that violate the Twitter Rules. Most of the accounts we suspend are suspended because
they are automated or fake accounts, and they introduce security risks for Twitter and all of the
individuals who use our service. These types of accounts are contrary to our Twitter Rules.
Unfortunately, sometimes a real person’s account gets suspended by mistake, and in those cases
we work with the person to make sure the account is unsuspended. If we suspect an account has
12
been hacked or compromised, we may suspend it until it can be secured and restored to the
account owner in order to reduce potentially malicious activity caused by the compromise.
As described above, we may suspend an account if it has been reported to us as violating
our Rules surrounding abuse. When an account engages in abusive behavior, like sending threats
to others, we may suspend it temporarily or, in some cases, permanently. An individual may be
able to unsuspend his or her own account by providing a phone number or confirming an email
address.
An account may also be temporarily disabled in response to reports of automated or
abusive behavior. For example, an individual may be prevented from Tweeting from his or her
account for a specific period of time or may be asked to verify certain information before
proceeding.
If an account was suspended or locked in error, an individual can appeal. First, the
individual must log in to the account that is suspended and file an appeal. The individual must
describe the nature of the appeal and provide an explanation of why the account is not in
violation of the Twitter Rules. Twitter employees will engage with the account holder via email
to resolve the suspension.
16. Mr. Dorsey, what specific steps does Twitter undertake to determine whether a Twitter
user’s account should be suspended?
Please see the answer to question 15.
17. Mr. Dorsey, what specific steps does Twitter undertake to determine whether a Twitter
user’ s account should be banned?
Please see the answer to question 15.
18. Mr. Dorsey, does Twitter only use content on the platform to make decisions about
suspending or removing Twitter accounts or does Twitter use data from outside entities to
make those decisions?
a. If yes, please identify the outside entities Twitter relies on for information
about Twitter users to make these decisions.
Please see the response to question five for additional information concerning our violent
extremist group policy and the circumstances under which we consider off-platform activity.
We take pride in Twitter being a platform where a diverse range of opinions can be held
and discussed, but we will not tolerate groups or individuals associated with them who engage in
and promote violence against civilians both on and off the platform. Accounts affiliated with
groups in which violence is a component of advancing their cause risk having a chilling effect on
opponents and bystanders. The violence that such groups promote could also have dangerous
13
consequences offline, jeopardizing their targets’ physical safety.
We prohibit the use of Twitter’s services by violent extremist groups – i.e., identified
groups subscribing to the use of violence as a means to advance their cause, whether political,
religious, or social. Groups that are prohibited on our services are those that identify as such or
engage in activity — both on and off the platform — that promotes violence. An individual on
Twitter may not affiliate with organizations that – whether by their own statements or activity
both on and off the platform – use or promote violence against civilians to further their causes.
We consider violent extremist groups to be those which identify through their stated
purpose, publications, or actions, as an extremist group; have engaged in, or currently engage in,
violence (and/or the promotion of violence) as a means to further their cause; and target civilians
in their acts (and/or promotion) of violence.
19. Mr. Dorsey, in late August, you were interviewed by CNN and you indicated “[Twitter
has] changed a lot. But [Twitter hasn’t] changed the underlying fundamentals.” Please
identify the underlying fundamentals of Twitter.
The mission we serve as Twitter, Inc. is to give everyone the power to create and share
ideas and information instantly without barriers. Our business will always follow that mission in
ways that improve – and do not detract from – a free and global conversation.
Twitter is the best place to see what’s happening and what people are talking about.
Every day, instances of breaking news, entertainment, sports, politics, big events and everyday
interests happen first on Twitter. Twitter is where the full story unfolds with live commentary
and where live events come to life unlike anywhere else. Our primary service can be accessed on
a variety of mobile devices, at twitter.com and via SMS.
20. Mr. Dorsey, in late August, you were interviewed by CNN and you indicated “We’ve
seen abuse, we’ve seen trolling, we’ve seen harassment, we’ve seen misinformation.” Did
you not anticipate this potential, either when you launched the platform or at any other
point since launching?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
14
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
21. Mr. Dorsey, as the company grew, were there ever discussions about implementing
greater security and enhanced transparency measures?
a. If yes, who was involved in those conversations and, if no, why not?
Discussions regarding security and transparency have occurred since the founding of
Twitter and are ongoing. We have dedicated Information Security and Platform teams who
focused on these issues as part of their core missions. Twitter recommends to the individuals on
its platform certain best security practices in order to help keep their accounts secure. These
include the use of a strong password that is not reused on other websites, the use login
verification, and requiring email and phone number to request a reset password link or code.
Twitter also suggests that individuals on the platform be cautious of suspicious links and always
make sure an individualis on twitter.com before he or she enters login information. We caution
people to never give their username and password out to third parties, especially those promising
to grow followers, make money, or verify an account.
Our biannual Twitter Transparency Report highlights trends in legal requests, intellectual
property-related requests, and email privacy best practices. The report also provides insight into
whether or not we take action on these requests. The Transparency Report includes information
requests from worldwide government and non-government legal requests we have received for
account information. Removal requests are also included in the Transparency Report and include
worldwide legal demands from governments and other authorized reporters, as well as reports
based on local laws from trusted reporters and non-governmental organizations, to remove or
withhold content.
The purpose of Twitter is to serve the public conversation, and we do not make value
judgments on personal beliefs. We are focused on making our platform—and the technology it
relies upon—better and smarter over time and sharing our work and progress with this
Committee and the American people. We think increased transparency is critical to promoting
healthy public conversation on Twitter and earning trust.
22. Mr. Dorsey, Twitter agreed to a “code of conduct” on Hate Speech under threat from
15
the European Union. Has it changed any of its policies as it affects the U.S. based on this
agreement, and has Twitter ever removed any U.S. based accounts on behalf of any foreign
government?
Twitter’s mission is to give everyone the power to create and share ideas and
information, and to express their opinions and beliefs without barriers. Free expression is a
human right – we believe that everyone has a voice, and the right to use it. Our role is to serve
the public conversation, which requires representation of a diverse range of perspectives.
We recognize that if people experience abuse on Twitter, it can jeopardize their ability to
express themselves. Research has shown that some groups of people are disproportionately
targeted with abuse online. This includes; women, people of color, lesbian, gay, bisexual,
transgender, queer, intersex, asexual individuals, marginalized and historically underrepresented
communities. For those who identity with multiple underrepresented groups, abuse may be more
common, more severe in nature and have a higher impact on those targeted.
We are committed to combating abuse motivated by hatred, prejudice or intolerance,
particularly abuse that seeks to silence the voices of those who have been historically
marginalized. For this reason, we prohibit behavior that targets individuals with abuse based on
protected category.
An individual on the platform is not permitted to promote violence against or directly
attack or threaten other people on the basis of race, ethnicity, national origin, sexual orientation,
gender, gender identity, religious affiliation, age, disability, or serious disease. We also do not
allow accounts whose primary purpose is inciting harm towards others on the basis of these
categories.
We do not allow individuals to use hateful images or symbols in your profile image or
profile header. Individuals on the platform as not allowed to use the username, display name, or
profile bio to engage in abusive behavior, such as targeted harassment or expressing hate towards
a person, group, or protected category.
Under this policy, we take action against behavior that targets individuals or an entire
protected category with hateful conduct. Targeting can happen in a number of ways, for
example, mentions, including a photo of an individual, referring to someone by their full name,
etc.
When determining the penalty for violating this policy, we consider a number of factors
including, but not limited to the severity of the violation and an individual’s previous record of
rule violations. For example, we may ask someone to remove the violating content and serve a
period of time in read-only mode before they can Tweet again. Subsequent violations will lead to
longer read-only periods and may eventually result in permanent account suspension. If an
account is engaging primarily in abusive behavior, or is deemed to have shared a violent threat,
we will permanently suspend the account upon initial review.
16
In regard to the removal of accounts, our biannual Twitter Transparency Report
highlights trends in legal requests, intellectual property-related requests, and email privacy best
practices. The report also provides insight into whether or not we take action on these requests.
The Transparency Report includes information requests from governments worldwide and
non-government legal requests we have received for account information. Removal requests are
also included in the Transparency Report and include worldwide legal demands from
governments and other authorized reporters, as well as reports based on local laws from trusted
reporters and non-governmental organizations, to remove or withhold content.
23. Mr. Dorsey, with the exclusion of the “Network of Enlightened Women,” which Twitter
appears to have added recently and have no record opposing online censorship, none of the
48 groups and individuals on the Twitter Trust and Safety Council lean even mildly
right-of-center. Does Twitter have plans to include more pro-free speech or conservative
voices in their product and policy decisions?
On Twitter, every voice has the power to shape the world. We see this power every day,
from activists who use Twitter to mobilize citizens to content creators who use Twitter to shape
opinion.
To ensure people can continue to express themselves freely and safely on Twitter, we
must provide more tools and policies. With hundreds of millions of Tweets sent per day, the
volume of content on Twitter is massive, which makes it extraordinarily complex to strike the
right balance between fighting abuse and speaking truth to power. It requires a multi-layered
approach where each individual on our platform has a part to play, as do the community of
experts working for safety and free expression.
The Twitter Trust and Safety Council provides input on our safety products, policies, and
programs. Twitter works with safety advocates, academics, and researchers; grassroots advocacy
organizations that rely on Twitter to build movements; and community groups working to
prevent abuse.
As we develop products, policies, and programs, our Trust & Safety Council help us tap
into the expertise and input of organizations at the intersection of these issues more efficiently
and quickly. Our diverse list of partners include safety advocates, academics, and researchers
focused on minors, media literacy, digital citizenship, and efforts around greater compassion and
empathy on the Internet; grassroots advocacy organizations that rely on Twitter to build
movements and momentum; and community groups with an acute need to prevent abuse,
harassment, and bullying, as well as mental health and suicide prevention.
These organizations are chosen based on their areas of focus and subject matter expertise,
rather than due to any particular political ideology. If the Chairman has additional suggestions
for groups that may be able to provide additional expertise on the Trust and Safety Council, we
welcome that input.
24. Mr. Dorsey, what percentage of Twitter content moderation reviews are conducted by
17
actual human beings rather than via automated processes and artificial intelligence?
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
25. Mr. Dorsey, how many employees have been hired and are currently assigned to
content moderation and reviews? Currently, what training and instruction do reviewers
receive to ensure Twitter users were in compliance with the company’s Terms of Service?
What salary range does Twitter pay employees assigned to content moderation and
reviews?
All individuals accessing or using Twitter’s services must adhere to the policies set forth
in the Twitter Rules. Failure to do so may result in Twitter taking one or more of the following
enforcement actions: (1) requiring an individual to delete prohibited content before he or she can
again create new posts and interact with other Twitter users; (2) temporarily limiting an
individual’s ability to create posts or interact with other Twitter users; (3) requesting an
individual verify account ownership with a phone number or email address; or (4) permanently
suspending an account or related accounts.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. In order to maintain a safe environment for
individuals on Twitter, we may suspend accounts that violate the Twitter Rules. Most of the
accounts we suspend are suspended because they are automated or fake accounts, and they
introduce security risks for Twitter and all of the individuals who use our service. These types of
accounts are contrary to our Twitter Rules. Unfortunately, sometimes a real person’s account
gets suspended by mistake, and in those cases we work with the person to make sure the account
is unsuspended. If we suspect an account has been hacked or compromised, we may suspend it
until it can be secured and restored to the account owner in order to reduce potentially malicious
activity caused by the compromise.
As described above, we may suspend an account if it has been reported to us as violating
our Rules surrounding abuse. When an account engages in abusive behavior, like sending threats
to others, we may suspend it temporarily or, in some cases, permanently. An individual may be
able to unsuspend his or her own account by providing a phone number or confirming an email
address.
An account may also be temporarily disabled in response to reports of automated or
abusive behavior. For example, an individual may be prevented from Tweeting from his or her
18
account for a specific period of time or may be asked to verify certain information before
proceeding.
If an account was suspended or locked in error, an individual can appeal. First, the
individual must log in to the account that is suspended and file an appeal. The individual must
describe the nature of the appeal and provide an explanation of why the account is not in
violation of the Twitter Rules. Twitter employees will engage with the account holder via email
to resolve the suspension.
26. Mr. Dorsey, does Twitter have a list of specific users it surveils more closely than other
Twitter users because of a history of use of fake accounts?
We have updated the Twitter Rules to provide clearer guidance around several key issues,
including fake account, attributed activity, and distribution of hacked materials. We have heard
feedback that people think our rules about spam and fake accounts only cover common spam
tactics like selling fake goods. As platform manipulation tactics continue to evolve, we are
updating and expanding our rules to better reflect how we identify fake accounts, and what types
of inauthentic activity violate our guidelines. We now may remove fake accounts engaged in a
variety of emergent, malicious behaviors. Some of the factors that we will take into account
when determining whether an account is fake include the use of stock or stolen avatar photos,
use of stolen or copied profile bios, and use of intentionally misleading profile information,
including profile location
Additionally, as per the Twitter Rules, if we are able to reliably attribute an account on
Twitter to an entity known to violate the Twitter Rules, we will take action on additional
accounts associated with that entity. We are expanding our enforcement approach to include
accounts that deliberately mimic or are intended to replace accounts we have previously
suspended for violating our rules. Further, our rules prohibit the distribution of hacked material
that contains private information or trade secrets, or could put people in harm’s way. We are also
expanding the criteria for when we will take action on accounts which claim responsibility for a
hack, which includes threats and public incentives to hack specific people and accounts.
Commentary about a hack or hacked materials, such as news articles discussing a hack, are
generally not considered a violation of this policy.
Twitter has increased our ability to identify and remove individuals who have previously
been banned from our service.
27. Mr. Dorsey, how do you define a “fake account” and what steps does Twitter under
take to identify fake accounts?
a. Once a fake account is identified, what specific steps does Twitter undertake
to verify that an account should be removed?
Twitter continues to develop the detection tools and systems needed to combat malicious
automation on our platform. Twitter has refined its detection systems. Twitter prioritizes
19
identifying suspicious account activity, such as exceptionally high-volume Tweeting with the
same hashtag or mentioning the same @handle without a reply from the account being
addressed, and requires an individual using the platform to confirm control. Twitter has also
increased its use of challenges intended to catch automated accounts, such as reCAPTCHAs, that
require individuals to identify portions of an image or type in words displayed on screen, and
password reset requests that protect potentially compromised accounts. Twitter is also in the
process of implementing mandatory email or cell phone verification for all new accounts.
Our efforts have been effective. Due to technology and process improvements, we are
now removing 214% more accounts year-over-year for violating our platform manipulation
policies. For example, over the course of the last several months, our systems identified and
challenged between 8.5 million and 10 million accounts each week suspected of misusing
automation or producing spam. Spam can be generally described as unsolicited, repeated actions
that negatively impact other people. This includes many forms of automated account interactions
and behaviors as well as attempts to mislead or deceive people. This constitutes more than three
times the 3.2 million we were catching in September 2017. We thwart 530,000 suspicious logins
a day, approximately double the amount of logins that we detected a year ago.
These technological improvements have brought about a corresponding reduction in the
number of spam reports from people on Twitter, evidence to us that our systems’ ability to
automatically detect more malicious accounts and potential bad faith actors than they did in the
past. We received approximately 25,000 such reports per day in March of this year; that number
decreased to 17,000 in August.
We also removed locked accounts from people’s follower counts, to ensure these figures
are more reliable. Accounts are locked when our systems detect unusual activity and force a
password change or other challenge. If the challenge has not been met or the password has not
been changed within a month, the account is locked, barring it from sending Tweets, Retweets or
liking posts from others.
28. Mr. Dorsey, in January of this year, the New York Times published an article entitled
“The Follower Factory” that revealed some Twitter users purchase followers and retweets
from a company who specializes in selling automated bot accounts. Is it a violation of the
Twitter User Agreement to purchase followers and retweets ?
a. Since the publication of the New York Times’ article “The Follower Factory”
what steps has Twitter taken to address the issue of users purchasing fake
accounts and retweets?
b. What are the harms and concerns about artificially inflating the number of
followers on social media by paying for it?
Twitter Rules specifically prohibit users from creating, purchasing, and selling fake
accounts. Twitter considers such activity as prohibited spam, which includes creating fake
accounts, account interactions, or impressions; selling, purchasing, or attempting to artificially
20
inflate account interactions (such as followers, Retweets, likes, etc.); and using or promoting
third-party services or applications that claim to provide user accounts with more followers,
Retweets, or likes (such as follower trains, sites promising “more followers fast,” or any other
site that offers to automatically add followers or engagements to your account or Tweets).
Twitter proactively enforces these rules, including by requiring users to confirm their
phone numbers or pass a reCAPTCHA challenge to confirm that the account is being operated
by a real person. On a weekly basis, Twitter challenges or suspends millions of suspicious or
potentially fake accounts, including upon registration and log in.
After the New York Times article was published, Twitter used techniques, including
pattern identification, to identify potential fake accounts, including those described in the article.
These techniques must constantly evolve because the characteristics of fake accounts evolve
constantly to evade detection. Twitter also ran searches to identify copycat accounts, or accounts
that post content copied from actual users. This process can also lead to identification of fake
accounts.
29. Mr. Dorsey, the Twitter Rules provide that “if voices are silenced because people are
afraid to speak up” then the underlying philosophy of free speech and expression means
little.
a. Do algorithms make the determination that certain content silences voices
and, if so, what factors are considered by the algorithm ?
b. Do humans make the determination that certain content silences voices and,
if so, what factors are considered?
Twitter does not use political ideology to make any decisions -- whether the decision is
made by algorithms or humans. Political ideology does not factor into decisions regarding
ranking content on our service or how we enforce our rules. We believe strongly in being
impartial, and we strive to enforce our rules impartially. We do not shadowban anyone based on
political ideology. In fact, from a simple business perspective and to serve the public
conversation, Twitter is incentivized to keep all voices on the platform.
30. Mr. Dorsey, the Twitter Rules specifically state that users may not use the platform for
“any unlawful purpose or in furtherance of illegal activities.” If Twitter determines content
is unlawful or in furtherance of illegal activities, does Twitter notify law enforcement?
We have well-established relationships with law enforcement agencies, particularly in the
arena of hostile foreign action, including the Federal Bureau of Investigation Foreign Influence
Task Force and the Department of Homeland Security’s Election Security Task Force. We look
forward to continued cooperation with them on these issues, as only they have access to
information critical to our joint efforts to stop bad faith actors.
21
Information sharing and collaboration are critical to Twitter’s success in preventing
hostile foreign actors from disrupting meaningful political conversations on the platform. The
threat we face requires extensive partnership and collaboration with our government partners and
industry peers. We each possess information the other does not have, and our combined
information is more powerful in combating these threats together.
31. Mr. Dorsey, has Twitter, or any third parties it engages, audited its input data and
ranking system to determine exactly how much bias each contributes to output bias in
resulting Twitter search results? With what frequency does Twitter or others does conduct
such audits or studies?
We want Twitter to provide a useful, relevant experience to all people using our service.
With hundreds of millions of Tweets per day on Twitter, we ha have invested heavily in building
systems that organize content on Twitter to show individuals using the platform the most the
relevant information for that individual first. We want to do the work for our customers to make
it a positive and informative experience. With 335 million people using Twitter every month in
dozens of languages and countless cultural contexts, we rely upon machine learning algorithms
to help us organize content by relevance.
To preserve the integrity of our platform and to safeguard our democracy, Twitter has
also employed technology to be more aggressive in detecting and minimizing the visibility of
certain types of abusive and manipulative behaviors on our platform. The algorithms we use to
do this work are tuned to prevent the circulation of Tweets that violate our Terms of Service,
including the malicious behavior we saw in the 2016 election, whether by nation states seeking
to manipulate the election or by other groups who seek to artificially amplify their Tweets.
Despite the success we are seeing with our use of algorithms to combat abuse,
manipulation, and bad faith actors, we recognize that even a model created without deliberate
bias may nevertheless result in biased outcomes. Bias can happen inadvertently due to many
factors, such as the quality of the data used to train our models. In addition to ensuring that we
are not deliberately biasing the algorithms, it is our responsibility to understand, measure, and
reduce these accidental biases. This is an extremely complex challenge in our industry, and
algorithmic fairness and fair machine learning are active and substantial research topics in the
machine learning community. The machine learning teams at Twitter are learning about these
techniques and developing a roadmap to ensure our present and future machine learning models
uphold a high standard when it comes to algorithmic fairness. We believe this is an important
step in ensuring fairness in how we operate and we also know that it’s critical that we be more
transparent about our efforts in this space.
32. Mr. Dorsey, what suggestions do you have for raising the awareness of Twitter users by
signaling bias in Twitter search and timeline results?
Twitter does not use political ideology to make any decisions -- whether related to
ranking content on our service or how we enforce our rules. We believe strongly in being
22
impartial, and we strive to enforce our rules impartially. We do not shadowban anyone based on
political ideology. In fact, from a simple business perspective and to serve the public
conversation, Twitter is incentivized to keep all voices on the platform.
33. Mr. Dorsey, Twitter announced August 30
th
a new issue ads policy and certification
process in the U.S. When will the company begin to enforce its new issue ads policy? Please
describe in detail what was the previous issue ads policy and certification process, if there
was one?
Twitter first implemented an updated Political Campaigning Policy to provide clearer
guidance about how we define political content and who can promote political content on our
platform. Under the revised policy, advertisers who wish to target the United States with federal
political campaigning advertisements are required to self-identify as such and certify that they
are located within the United States. Foreign nationals will not be permitted to serve political ads
to individuals who identify as being located in the United States.
Twitter accounts that wish to target the U.S. with federal political campaigning
advertisements must also comply with a strict set of requirements. Among other things, the
account’s profile photo, header photo, and website must be identical to the individual’s or
organization’s online presence. In addition, the advertiser must take steps to verify that the
address used to serve advertisements with content related to a federal political campaign is
genuine.
To further increase transparency and better educate those who access promoted content,
accounts serving ads with content related to a federal political campaign will now be visually
identified and contain a disclaimer. This feature will allow people to more easily identify federal
political campaign advertisements, quickly identify the identity of the account funding the
advertisement, and immediately tell whether it was authorized by the candidate.
In June, we launched the Ads Transparency Center, which is open to everyone on Twitter
and the general public, and currently focuses on electioneering communications. Twitter requires
extensive information disclosures of any account involved in federal electioneering
communications and provides specific information to the public via the Ads Transparency
Center, including:
• Purchases made by a specific account;
• All past and current ads served on the platform for a specific account;
• Targeting criteria and results for each advertisement;
• Number of views each advertisement received; and
• Certain billing information associated with the account.
23
These are meaningful steps that will enhance the Twitter experience and protect the
health of political conversations on the platform.
In addition,Twitter’s efforts to provide transparency continue with the launch of a
U.S.-specific Issue Ads Policy and certification process, which occurred prior to the 2018 U.S.
midterm elections. The new policy impacts advertisements that refer to an election or a clearly
identified candidate or advertisements that advocate for legislative issues of national importance.
To provide people with additional information about individuals or organizations promoting
issue ads, Twitter has established a process that verifies an advertiser’s identity and location
within the United States. These advertisements will also be included in the Ads Transparency
Center. We are also examining how to adopt political campaigning and issue ads policies
globally. We remain committed to continuing to improve and invest resources in this space.
34. Mr. Dorsey, how many advertisers -- either individuals, organizations, or campaigns --
have applied and received certification under Twitter’s new process?
As of November 19, 2018, 234 electioneering and 321 issue advertisers were certified in
the United States. Twitter serves the public conversation by promoting health and earning the
trust of the people who use our service. We cannot succeed unless the American people have
confidence in the integrity of the information found on the platform, especially with respect to
information relevant to elections and the democratic process. In addition to the advertising
policies described in the answer to question 33, and to promote transparency and assist our
stakeholders in identifying messages from elected officials and those who are running for office,
we have made a concerted effort to verify all major party candidates for both federal and key
state positions. Through verification – a blue badge that appears next to a person’s Twitter
handle throughout the platform – we let people know that accounts of public interest are the
authentic accounts (as opposed to impersonation or parody accounts).
In addition, we have developed a new U.S. election label to identify political candidates.
The label includes information about the office the candidate is running for, the state the office is
located in, and the district number, if applicable. Accounts of candidates who have qualified for
the general election and who are running for governor or for the U.S. Senate or House of
Representatives will display an icon of a government building. These new features are designed
to instill confidence that the content people are viewing is reliable and accurately reflects
candidates’ and elected officials’ positions and opinions.
35. Mr. Dorsey, please explain in detail specifically how will campaign and issue ads be
labeled or highlighted in a user’s timeline? How many campaign and issue ads are
currently labeled under the new policy and process? From their timeline can a user now
click on the promoted ads to immediately see information about the advertiser’s identity
and location, like a “Learn More” button?
Please see the response to question 33.
36. Mr. Dorsey, under Twitter’s August 30
th
issues ad policy, will both campaign and issue
24
ads be viewable in Twitter’s Ads Transparency Center? Please explain what specific detail
will be viewable to a user?
Yes, both campaign and issue ads are viewable in the Twitter’s Ads Transparency Center.
In June, we launched the Ads Transparency Center, which is open to everyone on Twitter and
the general public, and currently focuses on electioneering communications. Twitter requires
extensive information disclosures of any account involved in federal electioneering
communications and provides specific information to the public via the Ads Transparency
Center, including:
• Purchases made by a specific account;
• All past and current ads served on the platform for a specific account;
• Targeting criteria and results for each advertisement;
• Number of views each advertisement received; and
• Certain billing information associated with the account.
37. Until 1987, the Fairness Doctrine required broadcasters to provide a right of reply to
ensure the presentation of balanced views on issues of public importance, during a time
when broadcast was a dominant news source for most Americans. While the FCC repealed
the Fairness Doctrine, the principle of providing balanced perspectives is still important in
journalism. When amplifying certain topics on the platform, does Twitter seek to ensure
that it promotes a balanced variety of viewpoints on a particular issue? If so, what steps are
taken to ensure representation of a variety of viewpoints, including those that are less
popular with users?
Twitter’s purpose is to serve the public conversation. We are an American company that
serves our global audience by focusing on the people who use our service, and we put them first
in every step we take. Twitter is used as a global town square, where people from around the
world come together in an open and free exchange of ideas. We must be a trusted and healthy
place that supports free and open discussion.
Twitter has publicly committed to improving the collective health, openness, and civility
of public conversation on our platform. Twitter’s health is measured by how we help encourage
more healthy debate, conversations, and critical thinking, including exposure to diverse
perspectives. Conversely, abuse, malicious automation, and manipulation detracts from the
health of our platform. We are committed to hold ourselves publicly accountable towards
progress of our health initiative.
There are other guiding objectives we consider to be core to our company. We must
ensure that all voices can be heard. We must continue to make improvements to our service so
that everyone feels safe participating in the public conversation – whether they are speaking or
25
simply listening. And we must ensure that people can trust in the credibility of the conversation
and its participants.
Let me be clear about one important and foundational fact: Twitter does not use political
ideology to make any decisions, whether related to ranking content on our service or how we
enforce our rules. We believe strongly in being impartial, and we strive to enforce our rules
impartially. We do not shadowban anyone based on political ideology. In fact, from a simple
business perspective and to serve the public conversation, Twitter is incentivized to keep all
voices on the platform.
38. In 1996, 10 years before the original Twitter platform was released, Section 230 of the
Communications Decency Act was enacted, distinguishing interactive online platforms
from traditional publishers by setting up a safe harbor protecting them from the lawsuits
publishers may face over third party generated content. This was based on the premise that
interactive services are essentially neutral platforms, not exercising the full editorial
judgment wielded by a publisher. Although like the internet message boards that were
prevalent when Section 230 was enacted, Twitter gets much of its content from the users,
given the curating power of the Twitter algorithms, coupled with other efforts on the
company's part to proactively promote or de-prioritize particular content or users, are you
exercising editorial judgment?
It is important to note that like any other private citizen, Twitter has a First Amendment
right to free speech. Twitter speaks when it enforces our rules about what content should and
should not be allowed on our platform. For example, by prohibiting members of violent
extremist groups from using the platform, Twitter is exercising its First Amendment right to
declare its disapproval of content from, or the message of, such groups.
39. In Six4three, LLC v. Facebook, Inc.,
a case concerning alleged anticompetitive effects of
its content management practices, Facebook recently argued that it is a publisher and that
its editorial decisions are therefore protected by the First Amendment.
a. Would you characterize Twitter in the same way?
b. Should content moderation efforts that make value judgments about quality,
veracity, or tone, similar to decisions made by publishers, affect the
applicability of Section 230, which specifies that Internet platforms cannot be
treated as publishers?
Private companies do have First Amendment rights, including rights to free speech, and
these rights should be considered.
40. Earlier this year, the Allow States and Victims to Fight Online Sex Trafficking Act of
2017 (FOSTA) was signed into law. This legislation represented the first successful attempt
to amend Section 230 since it was enacted in 1996.
26
a. Has Twitter changed any of its procedures with regard to detecting and
removing sex trafficking content as a result of FOSTA?
b. Has the legislation impacted Twitter’s bottom line?
c. Does the Twitter user experience differ in other markets where internet
platforms are not protected with safe harbor legislation similar to Section
230?
Sex trafficking is an illegal act. An individual who uses Twitter may not use our service
for any unlawful purposes or in furtherance of illegal activities. By using Twitter, an individual
agrees to comply with all applicable laws governing his or her online conduct and content.
Additionally, Twitter prohibits the promotion of adult sexual content globally. This
policy applies to Twitter’s paid advertising products. Examples of adult sexual content include
pornography, escort services and prostitution, full and partial nudity, dating sites which focus on
facilitating sexual encounters or infidelity, dating sites in which money, goods or services are
exchanged in return for a date; penis enlargement products & services and breast enhancement
services, modeled clothing that is sexual in nature, mail order bride services, sex toys, and host
and hostess clubs. All advertisers must comply with this advertising policy and with all
applicable laws and regulations.
41. In your testimony, you indicated that Wall Street did not approve of Twitter's recent
actions to de-activate suspicious accounts, ultimately lowering the number of followers of
certain accounts. While Twitter can afford to do this in the short term, Twitter, as a
publicly held company, has a fiduciary duty to its shareholders. How do you expect the
shifted focus on the health of the conversation taking place on the platform, rather than
maximizing user engagement, to impact the company?
The purpose of Twitter is to serve the public conversation. We serve our global audience
by focusing on the needs of the people who use our service, and we put them first in every step
we take. We want to be a global town square, where people from around the world come
together in an open and free exchange of ideas. We must be a trusted and healthy place that
supports free and open democratic debate.
Twitter is committed to improving the collective health, openness, and civility of public
conversation on our platform. Twitter’s is built and measured by how we help encourage more
healthy debate, conversations, and critical thinking. Conversely, abuse, malicious automation,
and manipulation detracts from it. We are committing Twitter to hold ourselves publicly
accountable towards progress.
We do see the health initiative and objective as a long-term growth vector for the
company. Although it does have short-term implications on daily active users, we ultimately
believe that we are driving this towards a growth vector. We believe it is important to solve
issues around unhealthy behavior on the platform and better the experience more broadly on
27
Twitter. This is an extremely important initiative to us, not only for the experience of Twitter,
but we believe the long-term growth of the platform, and we are proud of our progress so far.
42. For some events or topics, Twitter features custom emojis that automatically appear
when a user types a certain hashtag, which is a form of promoting or prioritizing certain
content. How is it decided to allow a custom emoji for a hashtag, and who designs the
emojis? Can users apply for this service? Are the designs done in consultation with the
event coordinators or users actively promoting these hashtags?
On Twitter, advertisers can design custom branded emojis that are triggered when a
specific hashtag is used. We know people quickly move through their timeline and a Tweet
needs to stand out to get attention. We call this “stopping power” and branded emojis help
marketers achieve it. The amount of attention ads receive increases by almost 10% when branded
emojis are included in the ad.
When branded emojis are paired with a promoted video, the emotional connection and
interest in the ad increases six-fold, as people are more focused on the ad. Further, campaigns
with branded emojis extend a brand’s presence across Twitter in a way that is personal and
authentic to the brand. In fact, the median number of earned media generated is 5.3 million
Tweet impressions, representing a 420% increase compared to the earned media baseline.
Twitter also supports emojis that can catalyze conversations around important national or
international events.
43. The “trending topics” list is one important method Twitter uses to promote or prioritize
certain content. Is a topic chosen to be featured in this list strictly based on how many posts
are made on a specific hashtag within a period of time, or do any human decisions
intervene? Anecdotally, some users have observed that hashtags with fewer tweets
sometimes appear on trending lists while those with more tweets do not.
a. A “tailored trends” list is generated based on a user’s location and who they
follow on Twitter. How does Twitter use location in determining which
trends
or content to display to an individual user? What is the source of the location
data being used?
When individuals on Twitter log into their accounts, they have immediate access to a
range of tools and account settings to access, correct, limit, delete or modify the personal data
provided to Twitter and associated with the account, including public or private settings,
marketing preferences, and applications that can access their accounts. These data settings can be
used to better personalize the individual’s use of Twitter and allow him or her the opportunity to
make informed choices about whether Twitter collects certain data, how it is used, and how it is
shared.
28
For example, individuals can change the personalization and data settings for their
Twitter account, including:
● Whether interest-based advertisements are shown to an individual on and off the
Twitter platform;
● How Twitter personalizes an individual’s experience across devices;
● Whether Twitter collects and uses an individual’s precise location;
● Whether Twitter personalizes their experience based on places they have been;
and
● Whether Twitter keeps track of the websites where an individual sees Twitter
content.
An individual on Twitter can disable all personalization and data setting features with a
single master setting prominently located at the top of the screen.
44. Rep. Walberg stated in his question to you that ISPs have certain transparency
requirements that require the ISPs to disclose if they are altering traffic, and asked you if a
similar idea would be helpful in the tech industry. You replied, “That [transparency] is a
good idea and would help earn people’s trust." What should those transparency
requirements look like with regard to content moderation practices?
Twitter is committed to the open exchange of information. First published on July 2,
2012, our biannual Twitter Transparency Report highlights trends in legal requests, intellectual
property-related requests, and email privacy best practices. The report also provides insight into
whether or not we take action on these requests. The Transparency Report includes information
requests from worldwide government and non-government legal requests we have received for
account information. Removal requests are also included in the Transparency Report and include
worldwide legal demands from governments and other authorized reporters, as well as reports
based on local laws from trusted reporters and non-governmental organizations, to remove or
withhold content.
The Transparency Report also includes information on government requests to remove
content that may violate Twitter’s Terms of Service (TOS) under the following Twitter Rules
categories: abusive behavior, copyright, promotion of terrorism, and trademark. It does not
include legal demands, regardless of whether they resulted in a TOS violation, which will
continue to be published in our removal request section report. As we take an objective approach
to processing global TOS reports, the fact that the reporters in these cases happened to be
government officials had no bearing on whether any action was taken under our Rules. We
continue to look for ways to improve The Transparency Report.
45. Mr. Dorsey, understanding that in your testimony you stated that algorithmic bias is a
relatively “new problem” and that Twitter is “early” in its learning process in terms of
managing it, what research, development, and other tool deployment is Twitter currently
undertaking to understand, manage, and address algorithmic bias on its platform?
29
We want Twitter to provide a useful, relevant experience to all people using our service.
With hundreds of millions of Tweets per day on Twitter, we ha have invested heavily in building
systems that organize content on Twitter to show individuals using the platform the most the
relevant information for that individual first. We want to do the work for our customers to make
it a positive and informative experience. With 335 million people using Twitter every month in
dozens of languages and countless cultural contexts, we rely upon machine learning algorithms
to help us organize content by relevance.
To preserve the integrity of our platform and to safeguard our democracy, Twitter has
also employed technology to be more aggressive in detecting and minimizing the visibility of
certain types of abusive and manipulative behaviors on our platform. The algorithms we use to
do this work are tuned to prevent the circulation of Tweets that violate our Terms of Service,
including the malicious behavior we saw in the 2016 election, whether by nation states seeking
to manipulate the election or by other groups who seek to artificially amplify their Tweets.
Despite the success we are seeing with our use of algorithms to combat abuse,
manipulation, and bad faith actors, we recognize that even a model created without deliberate
bias may nevertheless result in biased outcomes. Bias can happen inadvertently due to many
factors, such as the quality of the data used to train our models. In addition to ensuring that we
are not deliberately biasing the algorithms, it is our responsibility to understand, measure, and
reduce these accidental biases. This is an extremely complex challenge in our industry, and
algorithmic fairness and fair machine learning are active and substantial research topics in the
machine learning community. The machine learning teams at Twitter are learning about these
techniques and developing a roadmap to ensure our present and future machine learning models
uphold a high standard when it comes to algorithmic fairness. We believe this is an important
step in ensuring fairness in how we operate and we also know that it’s critical that we be more
transparent about our efforts in this space.
46. Mr. Dorsey, in your testimony, you stated that Twitter reviewers look for indicators of
“fairness” and “impartiality” when reviewing algorithmic output to try and identify bias.
a. Please provide Twitter' s definition of “fairness” as it applies to Twitter's
oversight of its algorithms.
b. Please provide Twitter's definition of “impartiality” as it applies to Twitter's
oversight of its algorithms.
Twitter uses a range of behavioral signals to determine how Tweets are organized and
presented in the home timeline, conversations, and search based on relevance. Twitter relies on
behavioral signals—such as how accounts behave and react to one another—to identify content
that detracts from a healthy public conversation, such as spam and abuse. Unless we have
determined that a Tweet violates Twitter policies, it will remain on the platform, and is available
in our product. Where we have identified a Tweet as potentially detracting from healthy
conversation (e.g.
, as potentially abusive), it will only be available to view if you click on “Show
more replies” or choose to see everything in your search setting.
30
Some examples of behavioral signals we use, in combination with each other and a range
of other signals, to help identify this type of content include: an account with no confirmed email
address, simultaneous registration for multiple accounts, accounts that repeatedly Tweet and
mention accounts that do not follow them, or behavior that might indicate a coordinated attack.
Twitter is also examining how accounts are connected to those that violate our rules and how
they interact with each other. The accuracy of the algorithms developed from these behavioral
signals will continue to improve over time.
These behavioral signals are an important factor in how Twitter organizes and presents
content in communal areas like conversation and search. Our primary goal is to ensure that
relevant content and Tweets contributing to healthy conversation will appear first in
conversations and search. Because our service operates in dozens of languages and hundreds of
cultural contexts around the globe, we have found that behavior is a strong signal that helps us
identify bad faith actors on our platform.
The behavioral ranking that Twitter utilizes does not consider in any way political views
or ideology. It focuses solely on the behavior of all accounts. Twitter is always working to
improve our behavior-based ranking models such that their breadth and accuracy will improve
over time. We use thousands of behavioral signals in our behavior-based ranking models—this
ensures that no one signal drives the ranking outcomes and protects against malicious attempts to
manipulate our ranking systems.
31
The Honorable Michael C. Burgess
1. Mr. Dorsey, you stated during the hearing that Twitter is building technologies so that it
doesn’t have to wait on reports from victims to act on violent or threatening content.
a. When do you anticipate implementation of these new technologies?
b. Currently, how many reports of inappropriate or threatening content does
Twitter receive per day?
i. How many of these reports result in action?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
In the cases of violent threats, Twitter recommends that in addition to reporting the
abusive content to the platform, the individual considers contacting local law enforcement and
we provide a tool that allows the individual to generate an email with all of the relevant
necessary information to submit a law enforcement report. Local law enforcement agencies can
accurately assess the validity of the threat, investigate the source of the threat, and respond to
concerns about physical safety. If Twitter is contacted by law enforcement directly, we can work
with them and provide the necessary information for their investigation of the threat. We
continuously deploy new technological tools to identify Tweets and accounts that violate our
Terms of Service.
32
2. Mistakes do happen, and we understand that, but we would like to understand what the
process is at Twitter if a mistake happens. That process is still very unclear.
a. If a user does not believe they have violated the Twitter Rules, what recourse
does that user have? What is the internal process at Twitter to review an
appeal? Are employees deciding whether they agree with a users ' posts?
b. There are high profile Twitter users who seem to have a different level of
recourse from, for example, a constituent of mine who uses Twitter for news
and conversations with friends. Are there different channels for review in
these cases?
In order to maintain a safe environment for individuals on Twitter, we may suspend
accounts that violate the Twitter Rules. Most of the accounts we suspend are suspended because
they are automated or fake accounts, and they introduce security risks for Twitter and all of the
individuals who use our service. These types of accounts are contrary to our Twitter Rules.
Unfortunately, sometimes a real person’s account gets suspended by mistake, and in those cases
we work with the person to make sure the account is unsuspended. If we suspect an account has
been hacked or compromised, we may suspend it until it can be secured and restored to the
account owner in order to reduce potentially malicious activity caused by the compromise.
We may suspend an account if it has been reported to us as violating our Rules
surrounding abuse. When an account engages in abusive behavior, like sending threats to others,
we may suspend it temporarily or, in some cases, permanently. An individual may be able to
unsuspend his or her own account by providing a phone number or confirming an email address.
An account may also be temporarily disabled in response to reports of automated or
abusive behavior. For example, an individual may be prevented from Tweeting from his or her
account for a specific period of time or may be asked to verify certain information before
proceeding.
If an account was suspended or locked in error, an individual can appeal. First, the
individual must log in to the account that is suspended and file an appeal. The individual must
describe the nature of the appeal and provide an explanation of why the account is not in
violation of the Twitter Rules. Twitter employees will engage with the account holder via email
to resolve the suspension.
3. Mr. Dorsey, I understand that Twitter is conducting internal investigations that will
result in a transparency report.
a. When do you anticipate being able to share this information with the Energy
and Commerce Committee?
Twitter is committed to the open exchange of information. First published on July 2,
33
2012, our biannual Twitter Transparency Report highlights trends in legal requests, intellectual
property-related requests, and email privacy best practices. The report also provides insight into
whether or not we take action on these requests. The Transparency Report includes information
requests from worldwide government and non-government legal requests we have received for
account information. Removal requests are also included in the Transparency Report and include
worldwide legal demands from governments and other authorized reporters, as well as reports
based on local laws from trusted reporters and non-governmental organizations, to remove or
withhold content.
The Transparency Report also discloses information on third-party requests that compel
Twitter to remove content for legal reasons (“legal requests”) under our Country Withheld
Content (“CWC”) policy. Governments (including law enforcement agencies), organizations
chartered to combat discrimination, and lawyers representing individuals are among the many
complainants that submit legal requests included below. For example, we may receive a court
order requiring the removal of defamatory statements in a particular jurisdiction, or law
enforcement may ask us to remove prohibited content such as Nazi symbols in Germany.
In December 2017, Twitter updated its in-product messaging about withheld content to
better explain why content has been withheld. Subsequently, we began to differentiate between
legal demands (e.g., court orders) and reports based on local law(s) (e.g., reports alleging the
illegality of particular content in a certain country). To further increase transparency, this change
is also reflected in the report below.
The Transparency Report also includes information on government requests to remove
content that may violate Twitter’s Terms of Service (TOS) under the following Twitter Rules
categories: abusive behavior, copyright, promotion of terrorism, and trademark. It does not
include legal demands, regardless of whether they resulted in a TOS violation, which will
continue to be published in our removal request section report. As we take an objective approach
to processing global TOS reports, the fact that the reporters in these cases happened to be
government officials had no bearing on whether any action was taken under our Rules.
The Transparency Report also includes the total number of Digital Millennium Copyright
Act (DMCA) takedown notices and counter notices received for Twitter and Periscope content,
along with data about the top five copyright reporters across both platforms. The Vine app was
transitioned in January of 2017. Trademark notices include reports of alleged Trademark Policy
violations received for Twitter and Periscope.
4. Mr. Dorsey, I've long been concerned about the role the Internet and online platforms,
such as Google, Facebook, Pinterest and Twitter, have played in enabling access to deadly
and illegal controlled substances. Given the current opioid crisis, would you please describe
in detail what specific policies and procedures Twitter has in place to crack down on illegal
online sales and marketing of opioids and pharmaceutical drugs.
a. Does Twitter have policies and procedures in place to disable the ability to
use the Twitter search function for the sales and marketing of controlled
34
substances? If yes, please outline in detail those policies and procedures. If
no, does Twitter plan on implementing such measures and when?
b. Does Twitter have policies and procedures in place to report to Federal, State
or international law enforcement information Twitter receives indicating
that
an individual or organization is engaged in the sale or marketing of
controlled substances? If yes, please outline in detail those policies and
procedures. If no, does Twitter plan on implementing such measures and
when?
c. Has Twitter established a 24/7 point of contact with whom Federal, State or
international law enforcement can communicate directly if law enforcement
has information indicating that an individual or organization is engaged in
the sale or marketing of controlled substances? If you have, please outline in
detail what person or department is Twitter is responsible, and provide their
contact information. If you have not, are you planning on implementing such
measures and when?
d. Does Twitter have specific information in its “Help Center” for users to
report the sale or marketing of controlled substances on the platform, similar
to the help modules on “online abuse” and “self-harm and suicide”? If yes,
please outline in detail what they are. If no, does Twitter plan on
implementing such measures and when?
Twitter agrees that addiction to opioids and the overdoses that result are incredibly
serious. Twitter appreciates the Committee’s strong leadership on this public health crisis. Our
terms of service are strong against illegal opioid sales: An individual who uses Twitter may not
use our service for any unlawful purposes or in furtherance of illegal activities. By using Twitter,
an individual agrees to comply with all applicable laws governing his or her online conduct and
content.
We have well-established relationships with law enforcement agencies, and we look
forward to continued cooperation with them on these issues, as often they have access to
information critical to our joint efforts to stop bad faith actors. The threat we face requires
extensive partnership and collaboration with our government partners and industry peers. We
each possess information the other does not have, and our combined information is more
powerful in combating these threats together. We have continuous coverage to address reports
from law enforcement around the world and have a portal to swiftly handle law enforcement
requests rendered by appropriate legal process. Additional information about our policies and
procedures for law enforcement requests can be found here:
https://help.twitter.com/en/rules-and-policies/twitter-law-enforcement-support.
Twitter has participated in events convened by White House to address this public health
crisis, including the Opioid Summit and Be Best launch. We have participated in summit
35
organized by the Food and Drug Administration, with researchers, industry peers, and various
governmental stakeholders, including Drug Enforcement Agency, the Department of Justice, and
Health and Human Services. And we have hosted a series of events on opioids, including an
event on "Combating the Opioid Epidemic" at our Washington, D.C. Twitter offices with the
Surgeon General and Republican members of Congress.
Additionally, we provided a custom emoji to support the Drug Enforcements Efforts
around #TakebackDay. We tweeted from official Twitter handles to drive engagement and
awareness around the event designed to encourage people to safely dispose of unwanted
prescription drugs. Twitter also provides ongoing Twitter trainings to nongovernmental
organizations focused on recovery efforts, including Lilly's Place and Young People in
Recovery, to maximize their reach and effectiveness. We provide these stakeholders information
on their use the platform to help people in need, provide education around abuse and recovery,
and increase fundraising on Twitter.
36
The Honorable Robert E. Latta
1. Mr. Dorsey, in your written statement you indicated Twitter conducted an internal
analysis of the Members of Congress affected by the auto-suggest search issue, and that you
would make that information available to the Committee, if requested. Please provide the
internal analysis done by Twitter with respect to the auto-suggest search issue.
In July 2018, we publicly acknowledged that some accounts were not being
auto-suggested even when people were searching for their specific name. Our technology was
using a decision-making criteria that considered the behavior of people following these accounts.
Specifically, the followers of these accounts had a higher proportion of abusive behavior on the
platform.
This issue impacted 600,000 accounts across the globe. The vast majority of impacted
accounts were not political in nature. The issue impacted 53 accounts of politicians in the U.S.,
representing 0.00883 percent of total affected accounts. This subset of affected accounts includes
10 accounts of Republican Members of Congress. The remainder of impacted political accounts
relate to campaign activity and affected candidates across the political spectrum.
To be clear, this issue is limited solely to the search auto-suggestion function. The
accounts, their Tweets, and all surrounding conversation about those accounts remained
displayed in search results. Once identified, this issue was promptly resolved within 24 hours.
An analysis of accounts for Members of Congress that were affected by this search issue
demonstrates that there was no negative effect on the growth of their follower counts. To the
contrary, follower counts of impacted Members of Congress increased.
Twitter will not reveal the names of the impacted accounts due to privacy and security
concerns, but we can disclose information concerning the accounts of those who have publicly
discussed they were impacted.
We have enclosed information concerning follower count data displayed in graph format
for the official accounts of Congressmen Matt Gaetz, Jim Jordan, Mark Meadows, and Devin
Nunes. The graphs contain publicly available data on the follower counts for the three month
period surrounding the incident, which occurred in late July. The other impacted members of
Congress also saw their followers increase during this time period.
2. Mr. Dorsey, in your testimony you indicated that Twitter has made more than 30 policy
and product changes since the beginning of last year to improve health on your platform.
Please provide a complete accounting of all 30-plus policy and product changes made.
Twitter recently developed and launched more than 30 policy and product changes
designed to foster information integrity and protect the people who use our service from abuse
and malicious automation. Twitter has made a number of improvements specifically in
37
preparation for the 2018 U.S. midterm election. Described below are highlights of our election
integrity work, and we are happy to follow up with you in greater detail.
Twitter continues to develop the detection tools and systems needed to combat malicious
automation on our platform. Twitter has refined its detection systems. Twitter prioritizes
identifying suspicious account activity, such as exceptionally high-volume Tweeting with the
same hashtag or mentioning the same @handle without a reply from the account being
addressed, and requires an individual using the platform to confirm control. Twitter has also
increased its use of challenges intended to catch automated accounts, such as reCAPTCHAs, that
require individuals to identify portions of an image or type in words displayed on screen, and
password reset requests that protect potentially compromised accounts. Twitter is also in the
process of implementing mandatory email or cell phone verification for all new accounts.
Our improvements include important structural changes. We recently reorganized the
structure of the company to allow our valued employees greater durability, agility, invention, and
entrepreneurial drive. The reorganization simplified the way we work, and enabled all of us to
focus on health of our platform.
In particular, we have created an internal cross-functional analytical team whose mission
is to monitor site and platform integrity. Drawing on expertise across the company, the analytical
team can respond immediately to escalations of inauthentic, malicious automated or
human-coordinated activity on the platform. The team’s work enables us to better understand the
nature of the malicious activity and mitigate it more quickly.
Our cross-functional team has developed a political conversations dashboard to evaluate
the integrity of political conversations on the platform in the aggregate, focusing primarily (but
not exclusively) on elections in the United States in the near term. For example, this dashboard
surfaces information about sudden shifts in sentiment around a specific conversation, suggesting
a potential coordinated campaign of activity, as well as information about groups of potentially
linked accounts that are posting about the same topic.
Twitter serves the public conversation by promoting health and earning the trust of the
people who use our service. We cannot succeed unless the American people have confidence in
the integrity of the information found on the platform, especially with respect to information
relevant to elections and the democratic process. To promote transparency and assist our
stakeholders in identifying messages from elected officials and those who are running for office,
we have made a concerted effort to verify all major party candidates for both federal and key
state positions. Through verification – a blue badge that appears next to a person’s Twitter
handle throughout the platform – we let people know that accounts of public interest are the
authentic accounts (as opposed to impersonation or parody accounts).
In addition, we have developed a new U.S. election label to identify political candidates.
The label includes information about the office the candidate is running for, the state the office is
located in, and the district number, if applicable. Accounts of candidates who have qualified for
the general election and who are running for governor or for the U.S. Senate or House of
Representatives will display an icon of a government building. These new features are designed
38
to instill confidence that the content people are viewing is reliable and accurately reflects
candidates’ and elected officials’ positions and opinions.
We have devoted considerable resources to increasing transparency and promoting
accountability in the ads served to Twitter customers. Twitter first implemented an updated
Political Campaigning Policy to provide clearer guidance about how we define political content
and who can promote political content on our platform. Under the revised policy, advertisers
who wish to target the United States with federal political campaigning advertisements are
required to self-identify as such and certify that they are located within the United States.
Foreign nationals will not be permitted to serve political ads to individuals who identify as being
located in the United States.
Twitter accounts that wish to target the U.S. with federal political campaigning
advertisements must also comply with a strict set of requirements. Among other things, the
account’s profile photo, header photo, and website must be identical to the individual’s or
organization’s online presence. In addition, the advertiser must take steps to verify that the
address used to serve advertisements with content related to a federal political campaign is
genuine.
To further increase transparency and better educate those who access promoted content,
accounts serving ads with content related to a federal political campaign will now be visually
identified and contain a disclaimer. This feature will allow people to more easily identify federal
political campaign advertisements, quickly identify the identity of the account funding the
advertisement, and immediately tell whether it was authorized by the candidate.
In June, we launched the Ads Transparency Center, which is open to everyone on Twitter
and the general public, and currently focuses on electioneering communications. Twitter requires
extensive information disclosures of any account involved in federal electioneering
communications and provides specific information to the public via the Ads Transparency Center
In addition,we recently announced the next phase of our efforts to provide transparency
with the launch of a U.S.-specific Issue Ads Policy and certification process. The new policy
impacts advertisements that refer to an election or a clearly identified candidate or
advertisements that advocate for legislative issues of national importance. To provide people
with additional information about individuals or organizations promoting issue ads, Twitter has
established a process that verifies an advertiser’s identity and location within the United States.
These advertisements will also be included in the Ads Transparency Center. We are also
examining how to adopt political campaigning and issue ads policies globally. We remain
committed to continuing to improve and invest resources in this space.
To further address malicious automation and abuse on the platform, we have also recently
updated our developer policies, which govern the access and use of public Tweet data made
available to developers and other third parties through our application programming interfaces
(“APIs”).
39
The Honorable Cathy McMorris Rodgers
1. Your company is an immense supporter of net neutrality regulations being applied to
Internet Service Providers (ISP's) based on Title II of the Telecommunications Act written
in 1934. We can all agree to the basic principles of Net Neutrality and that there shouldn't
be blocking, throttling or discrimination by those companies but I also believe that Title II
is the wrong approach.
a. Do you believe ISPs and Edge Providers have a different relationship to their
customers? If so, do you believe that ISPs and Edge Providers have different
responsibilities regarding the throttling and blocking of content?
b. Can you identify who your customers are?
c. Do you believe that standard consensus rules of the road regarding ISP
throttling and blocking should apply only ISPs?
d. Would you agree to have your company adhere to the same regulations you
are advocating the ISP's live by? And if not, why?
We disagree with the decision by the Federal Communications Commission to weaken
net neutrality provisions. American democracy is built on freedom of expression. As Benjamin
Franklin said, “Freedom of speech is a principal pillar of a free government: When this support is
taken away, the constitution of a free society is dissolved.” Increasingly, Americans are going
online to enter the public sphere. According to a March 2017 poll, 57 percent of Americans
shared their political opinions or feelings via social media channels. Nearly 80 percent believe
social media channels impact public policy outcomes, and more than half said social media had
some impact on their voting decision.
Twitter allows every user to have a voice. As a global platform for free expression and
live conversation, Twitter has become a significant medium to give voice to the voiceless and to
speak truth to power. Our open platform facilitates conversations and connections that would
have been unlikely or impossible before the first Tweet was sent in 2006. From popular uprisings
that challenge oppressive regimes during the #ArabSpring to local tragedies that highlight social
injustice in #Ferguson, Missouri, these moments could not have emerged from local connections
to reach an international audience without the free and open conversation that Twitter facilitates.
Furthermore, Twitter has advanced how elected officials and candidates communicate
with constituents and the electorate, allowing them to connect in real-time on a global scale
without permission from large corporate gatekeepers. At least 856 Twitter accounts belong to
heads of state and government and foreign ministers in 178 countries, representing 92 percent of
all United Nations member states, with a combined audience of 356 million followers. Because
of platforms like Twitter, virtually anyone can share his or her views on issues of local or
national importance with government leaders. As a result, leaders have a stronger understanding
of both the people whom they represent and issues of civic interest as well as increased
40
accountability.
Net Neutrality is one of the most important free expression issues of our time. Without
Net Neutrality, Internet Service Providers could charge content providers gratuitous new fees to
reach Internet users and discriminate in the prioritized delivery of such content, frustrating the
free flow of information. Without Net Neutrality in force, ISPs would be able to block content
they don’t like, reject apps and content that compete with their own offerings, and arbitrarily
discriminate against content providers by prioritizing the Internet traffic of some over others.
ISPs and edge providers have a different relationships with their users. Because
broadband Internet access is increasingly provided by fewer and fewer companies in a
concentrated ISP marketplace, smaller and noncommercial voices may be impacted adversely
because they are unable to pay a broadband toll for “fast lane” service. Allowing ISPs to
determine what content is relegated to the backwaters of the Internet in second or third-tier status
could reduce the visibility and impact of important voices in the local, national, or global media
mix. In this context, Net Neutrality is a modern-day safeguard to protect freedom of expression.
Without strong rules supporting Net Neutrality, Twitter’s core value of promoting and
supporting the freedom of expression of our users is at risk.
2. On Privacy, there is clearly a significant public debate about private companies’ access
to, storage of, and monetization of people's data.
a. What do you believe is the role of consumers regarding control of their
information?
b. Do you believe that we need greater transparency of private companies'
privacy and data usage practices?
c. Do you believe consumers should have greater control over their
information? What do you make of proposed opt-in policies for Internet
Service Providers?
d. Do you believe ISPs and Edge Providers like you should be subject to the
same or different standards as it relates to consumer privacy? If so, can you
elaborate?
To ensure such trust, the privacy of the people who use our service is of paramount
importance to Twitter. We believe privacy is a fundamental right, not a privilege. Privacy is part
of Twitter’s DNA. Since Twitter’s creation over a decade ago, we have offered a range of ways
for people to control their experience on Twitter, from creating pseudonymous accounts to
letting people control who sees their Tweets, in addition to a wide array of granular privacy
controls. This deliberate design has allowed people around the world using Twitter to protect
their privacy.
41
That same philosophy guides how we work to protect the data people share with Twitter.
Twitter empowers the people who use our services to make informed decisions about the data
they share with us.
Twitter believes individuals should know, and have meaningful control over, what data is
being collected about them, how it is used, and when it is shared. Twitter is always working to
improve transparency into what data is collected and how it is used. Twitter designs its services
so that individuals can control the personal data that is shared through our services. People that
use our services have tools to help them control their data. For example, if an individual has
registered an account, through their account settings they can access, correct, delete or modify
the personal data associated with their account.
42
The Honorable David McKinley
1. A recent study in the American Journal of Public Health analyzed a number of tweets
over 5 months and identified nearly 2000 sites linked to illicit online drug sales on Twitter.
Your website states that this is prohibited. Can you tell me how many of those 2000 sites
are still up?
Twitter agrees that addiction to opioids and the overdoses that result are incredibly
serious. Twitter appreciates the Committee’s strong leadership on this public health crisis. Our
rules of service are strong on illegal opioid sales: An individual who uses Twitter may not use
our service for any unlawful purposes or in furtherance of illegal activities. By using Twitter, an
individual agrees to comply with all applicable laws governing his or her online conduct and
content. We remove any accounts that are linked to illicit online drug sales as soon as they are
brought to our attention. We continuously improve our ability to locate and remove accounts
claiming to sell drugs on our platform.
We have well-established relationships with law enforcement agencies, and we look
forward to continued cooperation with them on these issues, as often they have access to
information critical to our joint efforts to stop bad faith actors. The threat we face requires
extensive partnership and collaboration with our government partners and industry peers. We
each possess information the other does not have, and our combined information is more
powerful in combating these threats together.
Twitter has participated in events convened by White House to address this public health
crisis, including the Opioid Summit and Be Best launch. We have participated in summit
organized by the Food and Drug Administration, with researchers, industry peers, and various
governmental stakeholders, including Drug Enforcement Agency, the Department of Justice, and
Health and Human Services. And we have hosted a series of events on opioids, including an
event on "Combating the Opioid Epidemic" at our Washington, D.C. Twitter offices with the
Surgeon General and Republican members of Congress.
Additionally, we provided a custom emoji to support the Drug Enforcements Efforts
around #TakebackDay. We tweeted from official Twitter handles to drive engagement and
awareness around the event designed to encourage people to safely dispose of unwanted
prescription drugs. Twitter also provides ongoing Twitter trainings to nongovernmental
organizations focused on recovery efforts, including Lilly's Place and Young People in
Recovery, to maximize their reach and effectiveness. We provide these stakeholders information
on their use the platform to help people in need, provide education around abuse and recovery,
and increase fundraising on Twitter.
2. Can you explain Twitter’s current process for locating and removing criminal and
prohibited activity on the platform? What actions are you taking to make this process more
efficient and effective?
43
Anyone can report illegal behavior directly from a Tweet, profile, or Direct Message. An
individual navigates to the offending Tweet, account, or message and selects an icon that reports
that it is harmful. Multiple Tweets can be included in the same report, helping us gain better
context while investigating the issues to resolve them faster. For some types of report Twitter
also prompts the individual to provide more information concerning the issue that is being
reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. An individual can also contact local law enforcement
and we provide a tool that allows the individual to generate an email with all of the relevant
necessary information to submit a law enforcement report. Local law enforcement agencies can
accurately assess the validity of the threat, investigate the source of the threat, and respond to
concerns raised by the Tweet. If Twitter is contacted by law enforcement directly, we can work
with them and provide the necessary information for their investigation of the threat.
We also recognize that collaboration with our industry peers and civil society is critically
important to addressing common threats and that it has been successful in meeting shared
challenges. In June 2017, for example, we launched the Global Internet Forum to Counter
Terrorism (the “GIFCT”), a partnership among Twitter, YouTube, Facebook, and Microsoft.
The GIFCT facilitates, among other things, information sharing; technical cooperation;
and research collaboration, including with academic institutions. In September 2017, the
members of the GIFTC announced a multimillion dollar commitment to support research on
terrorist abuse of the Internet and how governments, tech companies, and civil society can
respond effectively. We are looking to establish a network of experts that can develop these
platform-agnostic research questions and analysis that consider a range of geopolitical contexts.
The GIFCT has created a shared industry database of “hashes”—unique digital
“fingerprints”—for violent terrorist imagery or terrorist recruitment videos or images that have
been removed from our individual services. The database allows a company that discovers
terrorist content on one of its sites to create a digital fingerprint and share it with the other
companies in the forum, who can then use those hashes to identify such content on their services
or platforms, review against their respective policies and individual rules, and remove matching
content as appropriate, or even block extremist content before it is posted in the first place.
The database now contains more than 88,000 hashes. Instagram, Justpaste.it, LinkedIn,
Oath, and Snap have also joined this initiative, and we are working to add several additional
companies in 2018. Twitter also participates in the Technology Coalition, which shares images
to counter child abuse. The database works to surface content for human review against each
platform’s respective terms of service. This is essential to take into account the context, for
example academic or news media use.
Because each platform is unique, there are many elements of our coordinated work that
do not translate easily across platforms. Although we share with other companies our approach
to addressing shared threats, including certain signals that we use to identify malicious content,
44
solutions applicable to the Twitter platform are not always applicable to other companies. We
describe our tools as “in-house and proprietary” to distinguish them from tools that are
developed by and licensed from third-party vendors.
45
The Honorable Gus Bilirakis
1. I have heard from my local school districts that they must consistently respond to threats
of school violence. In your response you mentioned that the company conducts outreach to
local entities and law enforcement officials when you see anything impacting potential
physical security threats. What are the specific implementations of that process?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
In the cases of violent threats, Twitter recommends that in addition to reporting the
abusive content to the platform, the individual considers contacting local law enforcement and
we provide a tool that allows the user to generate an email with all of the relevant necessary
information to submit a law enforcement report. Local law enforcement agencies can accurately
assess the validity of the threat, investigate the source of the threat, and respond to concerns
about physical safety. If Twitter is contacted by law enforcement directly, we can work with
them and provide the necessary information for their investigation of the threat. Twitter may also
contact law enforcement proactively if we become aware of an exigent situation that is likely to
result in imminent harm or death.
2. How does Twitter determine the efficacy of the company's process for responding to
potential threats?
46
All individuals accessing or using Twitter’s services must adhere to the policies set forth
in the Twitter Rules. Failure to do so may result in Twitter taking one or more of the following
enforcement actions: (1) requiring an individual to delete prohibited content before he or she can
again create new posts and interact with other Twitter users; (2) temporarily limiting an
individual’s ability to create posts or interact with other Twitter users; (3) requesting an
individual verify account ownership with a phone number or email address; or (4) permanently
suspending an account or related accounts.
Accounts under investigation or which have been detected as sharing content in violation
with the Twitter Rules may have their account or visibility limited in various parts of Twitter,
including search. Our policies and enforcement options evolve continuously to address emerging
behaviors online.
Our enforcement options have expanded significantly over the years. We originally had
only one enforcement option: account suspension. Since then, we’ve added a range of
enforcement actions and now have the ability to take action at the Tweet, Direct Message, and
account levels. Additionally, we take measures to educate individuals that have violated our rules
about the specific tweet(s) in violation and which policy has been violated. We also continue to
improve the technology we use to prioritize reports that are most likely to violate our rules and
last year we introduced smarter, more aggressive witness reporting to augment our approach.
We have embarked on an effort to better measure the health of the conversation on our
platform and will be transparent with our results.
3. You said you are always looking for ways to improve the process for quickly alerting the
company to potential threats and responding with outreach to the proper authorities, and
would be open to an implementation that can be scaled. In what ways would you consider
working directly with local school districts and institutions, which play a key role in
protecting its student body?
Twitter is continuing to explore and invest in what more we can do with our technology,
enforcement options, and policies – not just in the U.S., but to everyone we serve around the
world. We are committed to continuing to improve and to holding ourselves accountable as we
work to make Twitter better for everyone.
We are open to ideas, from local school districts and other institutions, to better use of
our technology to catch people working to evade a Twitter suspension and identifying malicious,
automated accounts, and more quickly activating our team to ensure a human review element
continues to be present within our all of automated processes.
Local school districts and other institutions should always raise any concerns to local law
enforcement. In line with our Privacy Policy, we may disclose account information to law
enforcement in response to a valid emergency disclosure request. Twitter evaluates emergency
disclosure requests on a case-by-case basis in compliance with relevant law (e.g., 18 U.S.C. §
2702(b)(8) and Section 8 Irish Data Protection 1988 and 2003). If we receive information that
47
provides us with a good faith belief that there is an exigent emergency involving the danger of
death or serious physical injury to a person, we may provide information necessary to prevent
that harm, if we have it. If there is an exigent emergency that involves the danger of death or
serious physical injury to a person that Twitter may have information necessary to prevent, law
enforcement officers can submit an emergency disclosure request through our Legal Request
Submissions site (the quickest and most efficient method).
4. You mention that Twitter takes a data-driven approach to arranging Moments, based
primarily on the amount of conversation happening on a particular topic or event, and
then afterwards going into "impartiality" to provide many differing perspectives. Are
there editorial standards the company has set for what makes a Moment, and if so, what
are they? If no standards, why not, and would you consider using an official standard?
Moments are curated stories showcasing the very best of what’s happening on Twitter.
The Moments guide on Twitter is customized to show a person on the platform current topics
that are popular or relevant, so an individual can discover what is unfolding on Twitter in an
instant.
Twitter has standards and policies for our curators that are intended to reduce bias, and
increase accuracy and quality. Additional information concerning these policies is located here:
https://help.twitter.com/en/rules-and-policies/twitter-moments-guidelines-and-principles.
An individual can navigate to the Moments tab to see customized Moments from Today.
An individual can select Moments categorized by News, Sports, Entertainment, Fun, and more.
When a person the platform sees a Moment he or she would like to explore, the individual can
click it to view the entire story. A person the platform can like a moment, or reply, Retweet, and
like the Tweet captured by the Moment.
People on the platform create their own individual Twitter Moments. Twitter Moments
are curated stories about what’s happening around the world—powered by Tweets. It’s easy to
individuals to create their own story with Twitter Moments. There are three ways an individual
can begin creating his or her own Moment via twitter.com. The person can access Moments
through the Moments tab, on the profile page, or through a Tweet detail. To get started, an
individual only needs is a title, description, Tweets, and a selected cover image.
5. Regarding algorithms used to flag malicious behavior, what is the “error” rate, meaning
how often are tweets overturned upon review by an employee content reviewer?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
48
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
In the cases of violent threats, Twitter recommends that in addition to reporting the
abusive content to the platform, the individual considers contacting local law enforcement and
we provide a tool that allows the individual to generate an email with all of the relevant
necessary information to submit a law enforcement report. Local law enforcement agencies can
accurately assess the validity of the threat, investigate the source of the threat, and respond to
concerns about physical safety. If Twitter is contacted by law enforcement directly, we can work
with them and provide the necessary information for their investigation of the threat.
6. How many employees do you have to review content and make decisions on policing
behavior? How long does it take on average to get an “error” overturned? Is there
notification to user during the steps of this process?
In order to maintain a safe environment for individuals on Twitter, we may suspend
accounts that violate the Twitter Rules. Most of the accounts we suspend are suspended because
they are automated or fake accounts, and they introduce security risks for Twitter and all of the
individuals who use our service. These types of accounts are contrary to our Twitter Rules.
Unfortunately, sometimes a real person’s account gets suspended by mistake, and in those cases
we work with the person to make sure the account is unsuspended. If we suspect an account has
been hacked or compromised, we may suspend it until it can be secured and restored to the
account owner in order to reduce potentially malicious activity caused by the compromise.
We may suspend an account if it has been reported to us as violating our Rules
surrounding abuse. When an account engages in abusive behavior, like sending threats to others,
we may suspend it temporarily or, in some cases, permanently. An individual may be able to
unsuspend his or her own account by providing a phone number or confirming an email address.
49
An account may also be temporarily disabled in response to reports of automated or
abusive behavior. For example, an individual may be prevented from Tweeting from his or her
account for a specific period of time or may be asked to verify certain information before
proceeding.
If an account was suspended or locked in error, an individual can appeal. First, the
individual must log in to the account that is suspended and file an appeal. The individual must
describe the nature of the appeal and provide an explanation of why the account is not in
violation of the Twitter Rules. Twitter employees will engage with the account holder via email
to resolve the suspension.
7. Is the content reviewing algorithm technology unique to Twitter or is this something that
is developing industry wide? If the former and it shows promise, is it something they are
willing to share with competitors to provide a uniform industry standard?
Although we cannot speak to the innovative technologies being employed at other companies,
Twitter has strengthened its information sharing with its industry peers, particularly to address
shared challenges such as violent extremism, child sexual exploitation, and state-sponsored
disinformation operations. Information sharing and collaboration are critical to Twitter’s success
in preventing hostile foreign actors from disrupting meaningful political conversations on the
platform. We recognize the value of inputs we receive from our industry peers about hostile
foreign actors. We have shared and remain committed to sharing information across platforms to
better understand and address the threat of hostile foreign interference with the electoral process.
50
The Honorable Susan Brooks
1. During the hearing, I talked about a bill of mine that passed into law in 2015, the OHS
Social Media Improvement Act. The bill created a working group of relevant stakeholders
to put their heads together to look at the impact social media has on preparedness,
response, and recovery. The working group, housed in OHS' Science and Technology
Directorate, also looks at how to counteract misinformation spread via Twitter during
disasters and how first responders and communities can effectively use social media during
times of crisis. This working group has to-date made 3 reports, which highlight countering
false information in disasters and emergencies, best practices for incorporating social
media into exercises, and how to operationalize social media for public safety. You said you
were not aware of this working group, but would be willing to consider these reports.
Below I have provided links to the reports. If you would like more information about the
group, please reach out to my Legislative Assistant, Mimi Strobel (202.225.2276 or
• April 2016 - "From Concept to Reality: Operationalizing Social Media for
Preparedness, Response and Recovery;"
(https://www.dhs.gov/sites/default/files/publications/SMWG From-Concept-to
Reality-Operationalizing-Social-Media-508.pdf);
• March 2017 - "Best Practices for Incorporating Social Media into Exercises;"
(https://www.dhs.gov/pubIication/best-practices-incorporating-social-mediaexercise
s); and,
• March 2018 - "Countering False Information on Social Media in Disasters and
Emergencies" (https://www.dhs.gov/sites/default/files/publications/SMWG
Countering-False_Info-Social-Media-Disasters-EmergenciesMar2018-508.pdf).
Thank you for this information. Twitter agrees that our speed and reach are critical in
disaster response. We work cooperatively with governments around the world to better use the
platform to keep their constituencies safe. Additional information regarding our Crisis Response
and Natural Emergency work can be found here:
https://about.twitter.com/content/dam/about-twitter/values/twitter-for-good/Twitter-Crisis-Respo
nse-One-Pager.pdf
51
The Honorable Earl "Buddy" Carter
1. Mr. Dorsey, you mentioned during my questioning that you wanted to reboot the
verification process. Can you please describe the steps that Twitter is taking to overhaul
that process? What will the new process look like? What is the timeline?
Verification was meant to authenticate identity and voice but it is interpreted as an
endorsement or an indicator of importance. Twitter recognizes that we have created this
confusion and need to resolve it. Beginning in November 2017, We have paused all general
verifications while we work to resolve this issue.
Verification has long been perceived as an endorsement. We gave verified accounts
visual prominence on the service which deepened this perception. We should have addressed this
earlier but did not prioritize the work as we should have. This perception became worse when we
opened up verification for public submissions and verified people who we in no way endorse.
Twitter is currently working on a new authentication and verification program. In the
meantime, we are not accepting any public submissions for verification and have introduced new
guidelines for the program.
2. Mr. Dorsey, you mentioned that there is a new behavioral trend regarding algorithms
and the illegal opioid sales present on your platform. You also mentioned that you need to
look at how your algorithms are determining when they see this activity and can take
action. Please describe how you will be reviewing your algorithms for their behavior and
what changes you will be implementing.
Twitter agrees that addiction to opioids and the overdoses that result are incredibly
serious. Twitter appreciates the Committee’s strong leadership on this public health crisis. Our
rules of service are strong on illegal opioid sales: An individual who uses Twitter may not use
our service for any unlawful purposes or in furtherance of illegal activities. By using Twitter, an
individual agrees to comply with all applicable laws governing his or her online conduct and
content.
We have well-established relationships with law enforcement agencies, and we look
forward to continued cooperation with them on these issues, as often they have access to
information critical to our joint efforts to stop bad faith actors. The threat we face requires
extensive partnership and collaboration with our government partners and industry peers. We
each possess information the other does not have, and our combined information is more
powerful in combating these threats together.
Twitter has participated in events convened by White House to address this public health
crisis, including the Opioid Summit and Be Best launch. We have participated in summit
organized by the Food and Drug Administration, with researchers, industry peers, and various
governmental stakeholders, including Drug Enforcement Agency, the Department of Justice, and
Health and Human Services. And we have hosted a series of events on opioids, including an
52
event on "Combating the Opioid Epidemic" at our Washington, D.C. Twitter offices with the
Surgeon General and Republican members of Congress.
Additionally, we provided a custom emoji to support the Drug Enforcements Efforts
around #TakebackDay. We tweeted from official Twitter handles to drive engagement and
awareness around the event designed to encourage people to safely dispose of unwanted
prescription drugs. Twitter also provides ongoing Twitter trainings to nongovernmental
organizations focused on recovery efforts, including Lilly's Place and Young People in
Recovery, to maximize their reach and effectiveness. We provide these stakeholders information
on their use the platform to help people in need, provide education around abuse and recovery,
and increase fundraising on Twitter.
3. Mr. Dorsey, intellectual property violations not only hurt those who directly develop this
content, but also those industries that have developed around them. Across Georgia,
thousands of well-paying jobs have been developed as a result of this growth. How is
Twitter taking action to address intellectual property theft or rebroadcasting to keep up
with changing practices? A quick search found numerous accounts that have been up for
years with thousands of followers that continuously tweet out these links. How has Twitter
not addressed these accounts and what steps will you take moving forward?
Twitter responds to copyright complaints submitted under the Digital Millennium
Copyright Act (“DMCA”). Section 512 of the DMCA outlines the statutory requirements
necessary for formally reporting copyright infringement, as well as providing instructions on
how an affected party can appeal a removal by submitting a compliant counter-notice.
Twitter will respond to reports of alleged copyright infringement, such as allegations
concerning the unauthorized use of a copyrighted image as a profile or header photo, allegations
concerning the unauthorized use of a copyrighted video or image uploaded through our media
hosting services, or Tweets containing links to allegedly infringing materials. Note that not all
unauthorized uses of copyrighted materials are infringements.
Twitter’s response to copyright complaints may include the removal or restriction of
access to allegedly infringing material. If we remove or restrict access to user content in response
to a copyright complaint, Twitter will make a good faith effort to contact the affected account
holder with information concerning the removal or restriction of access, including a full copy of
the complaint, along with instructions for filing a counter-notice.
53
The Honorable Frank Pallone, Jr.
1. At the hearing, I asked you about Twitter' s training and resources for content
moderation. You could not provide specific responses at the time. Please provide specific
responses to the following questions:
a. How many human content moderators does Twitter employ in the U.S. and how
much do they get paid?
b. How many hours of training is given to them to ensure consistency in their
decisions?
Twitter strives to provide an environment where people can feel free to express
themselves. If abusive behavior happens, Twitter wants to ensure that it is easy for people to
report it to us. In order to ensure that people feel safe expressing diverse opinions and beliefs,
Twitter prohibits behavior that crosses the line into abuse, including behavior that harasses,
intimidates, or uses fear to silence another’s voice.
Twitter has approximately 4,000 employees worldwide and at various times in their
work, all will focus on improving the health of the conversation on Twitter. Our team is
continuously trained, not only as we are onboarded, but particularly following changes to the
Twitter Terms of Service.
Anyone can report abusive behavior directly from a Tweet, profile, or Direct Message.
An individual navigates to the offending Tweet, account, or message and selects an icon that
reports that it is abusive or harmful. Other options are available, for example posting private
information or a violent threat. Multiple Tweets can be included in the same report, helping us
gain better context while investigating the issues to resolve them faster. For some types of report
Twitter also prompts the individual to provide more information concerning the issue that is
being reported.
Twitter uses a combination of machine learning and human review to adjudicate abuse
reports and whether they violate our rules. Context matters when evaluating abusive behavior
and determining appropriate enforcement actions. Factors we may take into consideration
include, but are not limited to whether: the behavior is targeted at an individual or group of
people; the report has been filed by the target of the abuse or a bystander; and the behavior is
newsworthy and in the legitimate public interest. Twitter subsequently provides follow-up
notifications to the individual that reports the abuse. We also provide recommendations for
additional actions that the individual can take to improve his or her Twitter experience, for
example using the block or mute feature.
All individuals accessing or using Twitter’s services must adhere to the policies set forth
in the Twitter Rules. Failure to do so may result in Twitter taking one or more of the following
enforcement actions: (1) requiring an individual to delete prohibited content before he or she can
again create new posts and interact with other Twitter users; (2) temporarily limiting an
54
individual’s ability to create posts or interact with other Twitter users; (3) requesting an
individual verify account ownership with a phone number or email address; or (4) permanently
suspending an account or related accounts.
Accounts under investigation or which have been detected as sharing content in violation
with the Twitter Rules may have their account or visibility limited in various parts of Twitter,
including search. Our policies and enforcement options evolve continuously to address emerging
behaviors online.
2. What steps is Twitter taking to improve the consistency of its enforcement and the
metrics that demonstrate improvement?
Our enforcement options have expanded significantly over the years. We originally had
only one enforcement option: account suspension. Since then, we’ve added a range of
enforcement actions and now have the ability to take action at the Tweet, Direct Message, and
account levels. Additionally, we take measures to educate individuals that have violated our rules
about the specific tweet(s) in violation and which policy has been violated. We also continue to
improve the technology we use to prioritize reports that are most likely to violate our rules and
last year we introduced smarter, more aggressive witness reporting to augment our approach.
We believe an important component of improving the health on Twitter is to measure the
health of conversation that occurs on the platform. This is because in order to improve
something, one must be able to measure it. By measuring our contribution to the overall health of
the public conversation, we believe we can more holistically approach our impact on the world
for years to come.
Earlier this year, Twitter began collaborating with the non-profit research center Cortico
and the Massachusetts Institute of Technology Media Lab on exploring how to measure aspects
of the health of the public sphere. As a starting point, Cortico proposed an initial set of health
indicators for the United States (with the potential to expand to other nations), which are aligned
with four principles of a healthy public sphere. Those include:
● Shared Attention: Is there overlap in what we are talking about?
● Shared Reality: Are we using the same facts?
● Variety: Are we exposed to different opinions grounded in shared reality?
● Receptivity: Are we open, civil, and listening to different opinions?
Twitter strongly agrees that there must be a commitment to a rigorous and independently
vetted set of metrics to measure the health of public conversation on Twitter. And in order to
develop those health metrics for Twitter, we issued a request for proposal to outside experts for
their submissions on proposed health metrics, and methods for capturing, measuring, evaluating
and reporting on such metrics. Our expectation is that successful projects will produce
55
peer-reviewed, publicly available, open-access research articles and open source software
whenever possible.
As a result of our request for proposal, we are partnering with experts at the University of
Oxford and Leiden University and other academic institutions to better measure the health of
Twitter, focusing on informational echo chambers and unhealthy discourse on Twitter. This
collaboration will also enable us to study how exposure to a variety of perspectives and opinions
serves to reduce overall prejudice and discrimination. While looking at political discussions,
these projects do not focus on any particular ideological group and the outcomes will be
published in full in due course for further discussion.
56
The Honorable Debbie Dingell
1. What is the minimum number of individuals you allow to be targeted for
advertisements?
Tailored audiences is a product feature that advertisers use to target existing people on
the platform and customers. For example, Advertisers can reach existing customers by uploading
a list of email addresses. Advertisers can also target those that have recently visited the
advertisers’ websites or reach those that have taken specific action in an application, such as
installation or registration.
Twitter informs individuals on the platform about Tailored Audiences in several ways.
For example, Twitter describes this activity in its Privacy Policy, an “Ads info” footer on
twitter.com, and the “Why am I seeing this ad?” section of the drop down menu on Twitter ads
themselves. Each of these locations describe interest-based advertising on Twitter and explain
how to use the associated privacy controls. In addition, the “Your Twitter Data” tool allows
individuals on Twitter to download a list of advertisers that have included them in a Tailored
Audience.
If people on Twitter do not want Twitter to show them Tailored Audience ads on and off
of Twitter, there are several ways they can turn off this feature: using their Twitter settings, they
can visit the Personalization and data settings and adjust the Personalize ads setting; if they are
on the web, they can visit the Digital Advertising Alliance’s consumer choice tool at
optout.aboutads.info to opt out of seeing interest-based advertising from Twitter in their current
browser; if they do not want Twitter to show them interest-based ads in Twitter for iOS on their
current mobile device, they can enable the “Limit Ad Tracking” setting in their iOS phone’s
settings; and if they do not want Twitter to show them interest-based ads in Twitter for Android
on their current mobile device, they can enable “Opt out of Ads Personalization” in an Android
phone’s settings.
In addition to explaining Tailored Audiences to people on the platform, offering them
several ways to disable the feature, and enabling them to view the advertisers who have included
them in Tailored Audiences, as described above, the Tailored Audience legal terms require that
advertisers have secured all necessary rights, consents, waivers, and licenses for use of data.
Advertisers are also required to provide all people from whom the data is collected with
legally-sufficient notice that fully discloses the collection, use, and sharing of the data that is
provided to Twitter for purposes of serving ads targeted to people’s interest, and legally
sufficient instructions on how they can opt out of interest-based advertising on Twitter.
2. Do you have any existing partnerships with device makers?
Twitter does have agreements in place with a wide range of device manufacturers around
the world, which grant rights to pre-install applications to use Twitter on mobile devices. These
agreements are meant to eliminate the need for end-users to manually go through the download
57
and installation process upon device purchase and activation; they do not otherwise grant access
to Twitter user information. They provide a user experience similar to that in the United States,
where consumers see various applications pre-populated on their phones when initializing
service.
Twitter also has agreements in place to enable device manufacturers to provide a “Find
Your Friends”-like feature on mobile phones, which allows users to find the Twitter handles of
their existing contacts. This feature takes a user’s contact list (email addresses and phone
numbers) and returns associated Twitter handles via an API. This integration does not return or
provide email addresses or phone numbers. Like within the Twitter application, this feature
respects user choice: any Twitter user can turn off their discoverability setting, and they will not
be discoverable through the application program interface, or API.
It is important to note that all Twitter APIs honor the long-standing principle of
defending the privacy of our users, reflecting our core values as a company. As stated in our
Developer Agreement and Policy, we prohibit the use of Twitter content for targeting,
segmenting, or profiling individuals based on health, negative financial status or condition,
political affiliation or beliefs, race, ethnicity, religious or philosophical affiliation or beliefs, sex
life or sexual orientation, trade union membership, data relating to any alleged or actual
commission of a crime, or any other sensitive categories of personal information prohibited by
law.
We also prohibit any entity accessing Twitter content from conducting any analyses or
research that isolates a group of individuals or any single individual for any unlawful or
discriminatory purpose or in a manner that would be inconsistent with our users’ reasonable
expectations of privacy.
While Twitter does not sell private user data, we note that users of our commercial data
platform seeking access to public data must complete a rigorous review and approval process
before we grant them access to Twitter data via our enterprise and premium APIs, and are
subject to regular reviews and policy checks once they have access. API users who are found to
be in violation of our policies are subject to enforcement actions, including immediate
termination of API access.
3. When you look at the performance of your AI, what is an acceptable error rate?
We want Twitter to provide a useful, relevant experience to all people using our service.
With hundreds of millions of Tweets per day on Twitter, we ha have invested heavily in building
systems that organize content on Twitter to show individuals using the platform the most the
relevant information for that individual first. We want to do the work for our customers to make
it a positive and informative experience. With 335 million people using Twitter every month in
dozens of languages and countless cultural contexts, we rely upon machine learning algorithms
to help us organize content by relevance.
Twitter is continuing to improve our systems so they can better detect any issues that
58
arise and correct for them. We will continue to refine our approach and will be transparent about
why we make the decisions that we do.
American technology companies have thrived, in part, because we have had the freedom
to experiment, learn and iterate at great speed. We are continuously improving our machine
learning tools, but an expectation that we will not, or cannot make mistakes, is unrealistic. We
will continue to tinker and learn from errors and will strive to be as transparent as possible along
the way.
4. Will you be deploying any of these Al programs to combat election propaganda?
Yes. To preserve the integrity of our platform and to safeguard our democracy, Twitter
has also employed algorithmic technology to be more aggressive in detecting and minimizing the
visibility of certain types of abusive and manipulative behaviors on our platform. The algorithms
we use to do this work are tuned to prevent the circulation of Tweets that violate our Terms of
Service, including the malicious behavior we saw in the 2016 election, whether by nation states
seeking to manipulate the election or by other groups who seek to artificially amplify their
Tweets. Additional information regarding our election integrity efforts can be found here:
https://about.twitter.com/en_us/values/elections-integrity.html
5. Because influence operations are always changing tactics, how often do you retrain your
models to adapt to these changes?
We continuously improve our approaches to combat malicious automation and curtail
efforts to manipulate the conversation on Twitter. Twitter uses a range of behavioral signals to
determine how Tweets are organized and presented in the home timeline, conversations, and
search based on relevance. Twitter relies on behavioral signals—such as how accounts behave
and react to one another—to identify content that detracts from a healthy public conversation,
such as spam and abuse. Unless we have determined that a Tweet violates Twitter policies, it will
remain on the platform, and is available in our product. Where we have identified a Tweet as
potentially detracting from healthy conversation (e.g.
, as potentially abusive), it will only be
available to view if you click on “Show more replies” or choose to see everything in your search
setting.
Some examples of behavioral signals we use, in combination with each other and a range
of other signals, to help identify this type of content include: an account with no confirmed email
address, simultaneous registration for multiple accounts, accounts that repeatedly Tweet and
mention accounts that do not follow them, or behavior that might indicate a coordinated attack.
Twitter is also examining how accounts are connected to those that violate our rules and how
they interact with each other. The accuracy of the algorithms developed from these behavioral
signals will continue to improve over time.
6. Does Twitter have the capacity to monitor IP addresses that access multiple accounts to
look for networks of malicious activity?
59
Twitter continues to develop the detection tools and systems needed to combat malicious
automation on our platform. Twitter has refined its detection systems. Twitter prioritizes
identifying suspicious account activity, such as exceptionally high-volume Tweeting with the
same hashtag or mentioning the same @handle without a reply from the account being
addressed, and requires an individual using the platform to confirm control. Twitter has also
increased its use of challenges intended to catch automated accounts, such as reCAPTCHAs, that
require individuals to identify portions of an image or type in words displayed on screen, and
password reset requests that protect potentially compromised accounts. Twitter is also in the
process of implementing mandatory email or cell phone verification for all new accounts.
IP addresses can be an important signal for our automated technology, but there are also
many other signal that can be useful, such as phone numbers and email addresses.
7. Like you, I believe understanding how AI comes to certain decisions is vital to building
public trust. Although the study of this is in its infancy, what work is Twitter doing to move
towards explainable AI?
This is an extremely complex challenge in our industry, and algorithmic fairness and fair
machine learning are active and substantial research topics in the machine learning community.
The machine learning teams at Twitter are developing a roadmap to ensure our present and
future machine learning models uphold a high standard when it comes to algorithmic fairness.
We believe this is an important step in ensuring fairness in how we operate and we also know
that it’s critical that we be more transparent about our efforts in this space.
60
The Honorable Jerry McNerney
1. As I noted at the hearing, some social media platforms have recently been accused of
facilitating discriminatory advertising-such as housing and employment ads. I want to
know what Twitter is doing to identify and remedy any potential similar practices. During
our exchange at the hearing, you stated “[w]e do regular audits of how our ads are targeted
and how they're delivered and we work to make sure that we have fairness within them.”
a. Are these audits performed by Twitter employees or are they conducted by
third parties?
b. What role do algorithms have in carrying out these audits?
c. Does each audit examine an individual ad? If not, how many ads are made
part of each audit?
d. How often are these audits conducted?
e. When did Twitter begin conducting these audits?
f. When you said "fairness within them" what did you mean by this?
g. Do these audits specifically examine whether an ad is being targeted in a
discriminatory way?
h. Do these audits specifically examine whether any discriminatory effects may
have resulted from Twitter's own ad optimization process?
i. Are there any other steps that Twitter takes to determine whether ads are
being targeted by advertisers in a discriminatory way? Please explain in
detail what these steps entail.
j. Are there any other steps that Twitter takes to determine whether
discriminatory effects have occurred as a result of its own ad optimization
process? Please explain in detail what these steps entail.
k. What happens once an instance is identified in which an ad has been targeted
in a discriminatory way or ultimately delivered in a way that resulted in
discriminatory effects?
Twitter does not allow discriminatory advertising, such as in housing or employment, on
our platform. An individual who uses Twitter may not use our service for any unlawful purposes
or in furtherance of illegal activities. By using Twitter, an individual agrees to comply with all
applicable laws governing his or her online conduct and content. Because discrimination in
61
housing and employment is against the law, an individual using Twitter may not engage in this
illegal activity on our platform.
In addition to this general prohibition against illegal activities by all individuals using
Twitter, there are specific Twitter Ads Policies that apply to paid advertising products on the
platform and prohibit discrimination. Our Twitter Ads Policies, as well as our legal agreements
that govern the use of Twitter for advertising, expressly state that advertisers on Twitter are
responsible for their Twitter Ads. This means all advertisers must follow all applicable laws and
regulations, create honest ads, and advertise safely and respectfully. Our policies and agreements
with advertisers require all advertisers to follow the law. The Twitter Privacy Policy emphasizes
the specific prohibition on using our advertising service in discriminatory ways, stating: “our ads
policies prohibit advertisers from targeting ads based on categories that we consider sensitive or
are prohibited by law, such as race, religion, politics, sex life, or health.”
Twitter takes violations of our Twitter Ads policies, the Twitter Rules, and Terms of
Service seriously. We will examine reported violations and take appropriate action, which may
include removal of offending advertisements and advertisers from the Twitter Ads platform.
Additional information about our ads policies can be found here:
https://business.twitter.com/en/help/ads-policies/other-policy-requirements/policies-for-keyword
-targeting.html
2. Does Twitter have any protocols in place for what actions should be taken when it is
discovered that ads are being targeted in a discriminatory way or being delivered in a way
that ultimately results in discriminatory effects? If so, please provide a copy of the
documents that describe these protocols.
The Twitter Terms of Service, Twitter Master Service Agreement (for advertising),
Twitter Ads Policies, and Twitter Privacy Policy each prohibit discriminatory advertising
behavior, as described above. Twitter has additional rules for advertisers using Twitter’s tailored
audiences products. Tailored audiences is a product feature that advertisers use to target existing
people on the platform and customers. For example, advertisers can reach existing customers by
uploading a list of email addresses. Advertisers can also target those that have recently visited
the advertisers’ websites or reach those that have taken specific action in an application, such as
installation or registration.
In addition to explaining Tailored Audiences to people on the platform, offering them
several ways to disable the feature, and enabling them to view the advertisers who have included
them in Tailored Audiences, the Tailored Audience legal terms require that advertisers have
secured all necessary rights, consents, waivers, and licenses for use of data.
62
Advertisers are also required to provide all people from whom the data is collected with
legally-sufficient notice that fully discloses the collection, use, and sharing of the data that is
provided to Twitter for purposes of serving ads targeted to people’s interest, and legally
sufficient instructions on how they can opt out of interest-based advertising on Twitter.
Twitter prohibits the creation of tailored audiences based on any sensitive information,
which includes: alleged or actual commission of a crime; health; genetic and/or biometric data;
negative financial status or condition; racial or ethnic origin; religious or philosophical affiliation
or beliefs; or sex life.
3. During the hearing, you acknowledged that it is possible for discrimination to result
from how advertisers are able to target ads on your platform (specifically, that they are
able to establish criteria that includes and excludes categories of users). Yet, as the
company's CEO, you were unable to answer my question if Twitter has ever taken down an
ad because of potential discriminatory effects and instead told me that you would have to
follow up to get the information. It was incredibly troubling that you did not know the
answer to this question after acknowledging that discrimination is possible.
a. Has Twitter ever taken down any ads because they were being targeted in a
discriminatory way? If so, how many?
b. Has Twitter ever taken down any ads because of discriminatory effects that
resulted from Twitter' s own ad optimization process? If so, how many?
c. Has Twitter ever taken down any ads because the content of the ad was
discriminatory? If so, how many?
The Twitter Terms of Service, Ads Policies, and Privacy Policy each prohibit
discriminatory advertising behavior, as described in the responses to questions one and two.
There are additional rules that govern the uses of keyword targeting in the Twitter timeline.
Advertisers using keyword targeting in timeline may not select keywords that target sensitive
categories.
Twitter prohibits keyword targeting based on any sensitive information, which includes:
alleged or actual commission of a crime; health; genetic and/or biometric data; negative financial
status or condition; racial or ethnic origin; religious or philosophical affiliation or beliefs; or sex
life.
Further, advertisers may not create advertisements which assert or imply knowledge of
personally identifiable or sensitive information, even when the ad has been created and targeted
63
without using such information. Advertisers using keyword targeting must adhere to this policy
and all other Twitter Ads policies in order to advertise.
Targeting people who use Twitter based on sensitive categories is a violation of our
Twitter Ads policies. Twitter takes violations of its Twitter Ads Policies, the Twitter Rules and
Terms of Service seriously. We will examine reported violations and take appropriate action,
which may include removal of offending advertisements and advertisers from the Twitter Ads
platform.
4. One reason that it is difficult for us to know if ads on Twitter' s platform are having
discriminatory effects is because there is no real way for watchdog groups to help identify
potential bias. When I asked you whether there is a way for watchdog groups to examine
how non-political ads are being targeted, you stated "[y]es, our Ads Transparency Center
is comprehensive of all ads." However, upon reviewing the information made available in
Twitter's Ads Transparency Center, it appears that no information is available about how
non-political ads are being targeted.
a. On what date do you plan to make available information about how
non-political ads are being targeted and who is seeing the ads?
b. Why does the Ads Transparency Center only include information about ads
during the last seven days? Do you plan to extend this time period?
Twitter first implemented an updated Political Campaigning Policy to provide clearer
guidance about how we define political content and who can promote political content on our
platform. Under the revised policy, advertisers who wish to target the United States with federal
political campaigning advertisements are required to self-identify as such and certify that they
are located within the United States. Foreign nationals will not be permitted to serve political ads
to individuals who identify as being located in the United States.
Twitter accounts that wish to target the U.S. with federal political campaigning
advertisements must also comply with a strict set of requirements. Among other things, the
account’s profile photo, header photo, and website must be identical to the individual’s or
organization’s online presence. In addition, the advertiser must take steps to verify that the
address used to serve advertisements with content related to a federal political campaign is
genuine.
To further increase transparency and better educate those who access promoted content,
accounts serving ads with content related to a federal political campaign will now be visually
identified and contain a disclaimer. This feature will allow people to more easily identify federal
political campaign advertisements, quickly identify the identity of the account funding the
advertisement, and immediately tell whether it was authorized by the candidate.
64
In June, we launched the Ads Transparency Center, which is open to everyone on Twitter
and the general public, and currently focuses on electioneering communications. Twitter requires
extensive information disclosures of any account involved in federal electioneering
communications and provides specific information to the public via the Ads Transparency
Center, including:
• Purchases made by a specific account;
• All past and current ads served on the platform for a specific account;
• Targeting criteria and results for each advertisement;
• Number of views each advertisement received; and
• Certain billing information associated with the account.
These are meaningful steps that will enhance the Twitter experience and protect the
health of political conversations on the platform.
In addition,we recently announced the next phase of our efforts to provide transparency
with the launch of a U.S.-specific Issue Ads Policy and certification process. The new policy
impacts advertisements that refer to an election or a clearly identified candidate or
advertisements that advocate for legislative issues of national importance. To provide people
with additional information about individuals or organizations promoting issue ads, Twitter has
established a process that verifies an advertiser’s identity and location within the United States.
These advertisements will also be included in the Ads Transparency Center. We are also
examining how to adopt political campaigning and issue ads policies globally. We remain
committed to continuing to improve and invest resources in this space.
For non-political ads, when individuals on Twitter log into their accounts, they have
immediate access to a range of tools and account settings to access, correct, limit, delete or
modify the personal data provided to Twitter and associated with the account, including public or
private settings, marketing preferences, and applications that can access their accounts. Twitter
also informs individuals on the platform about advertising in several ways. For example, Twitter
describes advertising activity in its Privacy Policy, an “Ads info” footer on twitter.com, and the
“Why am I seeing this ad?” section of the drop down menu on Twitter ads themselves.
5. When I asked you if Twitter is running any educational campaigns to inform users about
how their data is being used, you stated “[n]ot at the moment, but we should be looking at
that …”
a. What specific steps has Twitter taken since September 5, 2018 towards
running educational campaigns that let users know what information it
collects about them and how that information is used?
b. Has Twitter taken any other steps since September 5, 2018 to improve how it
65
informs users about what information is being collected and how that
c. Will you commit to running educational campaigns so that users are able to
better understand what information is being collected about them and how it
is used?
d. If so, on what date can we expect Twitter to launch these educational
campaigns?
Twitter’s purpose is to serve the public conversation. We serve our global audience by
focusing on the people who use our service, and we put them first in every step we take. People
around the world use Twitter as a “town square” to publicly, openly, and freely exchange ideas.
We must be a trusted and healthy place in order for this exchange of ideas and information to
continue.
To ensure such trust, the privacy of the people who use our service is of paramount
importance to Twitter. We believe privacy is a fundamental right, not a privilege. Privacy is part
of Twitter’s DNA. Since Twitter’s creation over a decade ago, we have offered a range of ways
for people to control their experience on Twitter, from creating pseudonymous accounts to
letting people control who sees their Tweets, in addition to a wide array of granular privacy
controls. This deliberate design has allowed people around the world using Twitter to protect
their privacy.
That same philosophy guides how we work to protect the data people share with Twitter.
Twitter empowers the people who use our services to make informed decisions about the data
they share with us. We believe individuals should know, and have meaningful control over, what
data is being collected about them, how it is used, and when it is shared.
Twitter recently updated our Privacy Policy to include callouts, graphics, and animations
designed to enable people to better understand the data we receive, how it is used, and when it is
shared.
6. Many consumers do not realize that Twitter makes inferences about them. Buried at the
bottom of one of its webpages titled “Your Twitter Data” is a link to “inferred interests
from Twitter.”
a. Aside from how ads are targeted, what are other ways in which “inferred
interests from Twitter” are used to personalize a user’s experience?
b. Is each user able to see all of the “inferred interests from Twitter” that have
been made about them?
c. If users are unable to see all of the “inferred interests from Twitter,” how
does Twitter determine which interests users are able to see?
66
d. When making these inferences, does Twitter take into account the user's
online activity on Twitter's platform as well online activity off of its
platform?
e. Will you commit to making available to users all of the “inferred interests
from Twitter” that have been made about them?
Although the information people share on Twitter is generally public, Twitter also
receives non-public personal information. For example, a person creating a Twitter account must
provide the platform with his or her email address or phone number. Twitter will also receive
standard log information, such as the device being used and the Internet Protocol (IP) address.
People who use the service may also choose to share additional information with Twitter
including, for example, their address book contacts in order to connect with people they know,
help others find and connect with them, and better recommend content to them and others.
In addition, and consistent with nearly all other online platforms, Twitter uses cookies
and other similar technologies, such as pixels or local storage, to operate its services and help
provide individuals on the platform with a better, faster, and safer experience. Cookies are small
files that websites place on a computer as an individual browses the web. Like many websites,
Twitter uses cookies to discover how people are using the services and to make them work
better. Twitter also uses cookies to help serve people more relevant content based on where they
have seen Twitter content on the web, and to serve targeted advertising. Twitter provides
individuals with additional control over whether their data is used for these purposes.
In order to show people the most interesting and relevant content, Twitter may infer
information about individuals based on their activity on the platform and other information. This
includes inferences such as what topics people may be interested in, how old a person is, what
languages a person speaks, and whether the content of one account may be of interest to others
on the platform. For example, Twitter may infer that an individual is a basketball fan based on
accounts the individual follows and suggest content related to the National Basketball
Association. Inferences assist Twitter in offering better services and personalizing the content
Twitter shows, including advertisements.
Twitter uses the data it receives to deliver, measure, and improve services in a variety of
ways, including: protecting the services; authentication and security; remembering preferences;
improving analytics and research, including Twitter Ads and Twitter buttons and widgets;
customizing Twitter services with more relevant content like tailored trends, stories,
advertisements, and suggestions for people to follow; and assisting in delivering advertisements,
measuring their performance, and making them more relevant.
Twitter informs individuals on the platform about Tailored Audiences in several ways.
For example, Twitter describes this activity in its Privacy Policy, an “Ads info” footer on
twitter.com, and the “Why am I seeing this ad?” section of the drop down menu on Twitter ads
themselves. Each of these locations describe interest-based advertising on Twitter and explain
67
how to use the associated privacy controls. In addition, the “Your Twitter Data” tool allows
individuals on Twitter to download a list of advertisers that have included them in a Tailored
Audience.
If people on Twitter do not want Twitter to show them Tailored Audience ads on and off
of Twitter, there are several ways they can turn off this feature: using their Twitter settings, they
can visit the Personalization and data settings and adjust the Personalize ads setting; if they are
on the web, they can visit the Digital Advertising Alliance’s consumer choice tool at
optout.aboutads.info to opt out of seeing interest-based advertising from Twitter in their current
browser; if they do not want Twitter to show them interest-based ads in Twitter for iOS on their
current mobile device, they can enable the “Limit Ad Tracking” setting in their iOS phone’s
settings; and if they do not want Twitter to show them interest-based ads in Twitter for Android
on their current mobile device, they can enable “Opt out of Ads Personalization” in an Android
phone’s settings.
In addition to explaining Tailored Audiences to people on the platform, offering them
several ways to disable the feature, and enabling them to view the advertisers who have included
them in Tailored Audiences, as described above, the Tailored Audience legal terms require that
advertisers have secured all necessary rights, consents, waivers, and licenses for use of data.
Advertisers are also required to provide all people from whom the data is collected with
legally-sufficient notice that fully discloses the collection, use, and sharing of the data that is
provided to Twitter for purposes of serving ads targeted to people’s interest, and legally
sufficient instructions on how they can opt out of interest-based advertising on Twitter.
7. Many consumers do not realize that Twitter collects “inferred interests from partners”
about them. This information is similarly buried at the very bottom of the webpage titled
“Your Twitter Data.”
a. Is each user able to view all of the “inferred interests from partners” that are
used to reach them on Twitter?
b. If users are unable to view all of the “inferred interests from partners,” how
does Twitter determine which ones users are able to see?
c. Will you commit to making available to users all of the “inferred interests
from partners”?
Twitter recently updated our Privacy Policy to include callouts, graphics, and animations
designed to enable people to better understand the data we receive, how it is used, and when it is
shared.
Twitter also provides a toolset called Your Twitter Data. Your Twitter Data tools provide
individuals accessible insights into the type of data stored by Twitter, such as username, email
address, and phone numbers associated with the account and account creation details. The
68
birthdays and locations of individuals are also shown in the tool if they have previously been
provided to Twitter.
Individuals using the Your Twitter Data tool can also see and modify certain information
that Twitter has inferred about the account and device such as gender, age range, languages, and
interests. People on Twitter can review inference information, advertisers who have included
them in tailored audiences, and demographic and interest data from external advertising partners.
The Your Twitter Data tool also allows people with a Twitter account to download a copy of
their relevant data from Twitter. We recently updated the download feature of the Your Twitter
Data tool to include additional information. Since that update on May 25, 2018, we have seen
approximately 586,000 people around the world use the tool to download 560 terabytes of data.
There is a version of this tool available to individuals who do not have a Twitter account,
or for those logged out of the account.
8. Many consumers do not realize that Twitter can track their activity across the web.
a. Is there a place on Twitter where users can go and see in plain text
information about the websites that the user has visited?
b. Will you commit to making this information available to users in plain text?
When individuals on Twitter log into their accounts, they have immediate access to a
range of tools and account settings to access, correct, limit, delete or modify the personal data
provided to Twitter and associated with the account, including public or private settings,
marketing preferences, and applications that can access their accounts. These data settings can be
used to better personalize the individual’s use of Twitter and allow him or her the opportunity to
make informed choices about whether Twitter collects certain data, how it is used, and how it is
shared.
For example, individuals can change the personalization and data settings for their
Twitter account, including:
● Whether interest-based advertisements are shown to an individual on and off the
Twitter platform;
● How Twitter personalizes an individual’s experience across devices;
● Whether Twitter collects and uses an individual’s precise location;
● Whether Twitter personalizes their experience based on places they have been;
and
● Whether Twitter keeps track of the websites where an individual sees Twitter
content.
An individual on Twitter can disable all personalization and data setting features with a
single master setting prominently located at the top of the screen.
69
People on the platform can also deactivate their accounts. Deactivated Twitter accounts,
including the display name, username, Tweets, and public profile information, are no longer
viewable on Twitter.com, Twitter for iOS, and Twitter for Android.
9. During the hearing, I asked whether Twitter stores previously collected data about users
after the user chooses to no longer have his or her activity tracked. You said “I believe it’s
erased, but we'll have to follow up with the details.”
a. If a user disables "Track where you see Twitter content across the web,"
does
Twitter still store any previously collected information?
I also asked you if you would commit to erasing data when users opt out of
the data being collected. You said “yes, but let just make sure I understand
the constraints and ramifications of that.” I want to make sure that you still
stand by your commitment.
b. Will you commit to giving users the option to have this information
permanently deleted?
Twitter believes individuals should know, and have meaningful control over, what data is
being collected about them, how it is used, and when it is shared. Twitter is always working to
improve transparency into what data is collected and how it is used. Twitter designs its services
so that individuals can control the personal data that is shared through our services. People that
use our services have tools to help them control their data. For example, if an individual has
registered an account, through their account settings they can access, correct, delete or modify
the personal data associated with their account.
10. It is my understanding that Twitter also collects information both on and off its
platform about non-users.
a. How are non-users able to see all of the information that Twitter collects
about them?
b. How are non-users able to delete all of the information that Twitter collects
about them?
Twitter believes individuals should know, and have meaningful control over, what data is
being collected about them, how it is used, and when it is shared. Twitter is always working to
improve transparency into what data is collected and how it is used. Twitter designs its services
so that individuals can control the personal data that is shared through our services. People that
use our services have tools to help them control their data. For example, if an individual has
registered an account, through their account settings they can access, correct, delete or modify
the personal data associated with their account.
70
Twitter recently updated our Privacy Policy to include callouts, graphics, and animations
designed to enable people to better understand the data we receive, how it is used, and when it is
shared.
Twitter also provides a toolset called Your Twitter Data. Your Twitter Data tools provide
individuals accessible insights into the type of data stored by Twitter, such as username, email
address, and phone numbers associated with the account and account creation details. The
birthdays and locations of individuals are also shown in the tool if they have previously been
provided to Twitter.
Individuals using the Your Twitter Data tool can also see and modify certain information
that Twitter has inferred about the account and device such as gender, age range, languages, and
interests. People on Twitter can review inference information, advertisers who have included
them in tailored audiences, and demographic and interest data from external advertising partners.
The Your Twitter Data tool also allows people with a Twitter account to download a copy of
their relevant data from Twitter. We recently updated the download feature of the Your Twitter
Data tool to include additional information. Since that update on May 25, 2018, we have seen
approximately 586,000 people around the world use the tool to download 560 terabytes of data.
There is a version of this tool available to individuals who do not have a Twitter account,
or for those logged out of the account. Instructions on how users with or without a Twitter
account can access this tool can be found in our help center here:
https://help.twitter.com/en/managing-your-account/accessing-your-twitter-data
71
The Honorable John Sarbanes
Media reports have indicated Twitter has become increasingly sensitive to allegations it
suffers from anti-conservative basis -- an allegation that continues to be advanced by many
conservatives. Some, including myself , have raised concern that Republican elected
officials and conservative leaders are intentionally drumming up accusations of political
bias at social media companies like Twitter in an effort to “work the refs,” securing
favorable treatment as recompense for non-existent bias.
As our Committee works to understand better the policies in place at your company to
prevent political bias, I am committed to better understanding how exactly those policies
are being developed. To that end, I greatly appreciate Twitter's thorough and candid
responses to the following questions.
1. It was reported in June that, in an effort to respond to complaints of bias against
conservatives on Twitter, you met with conservative leaders and Republican officials,
including Senator Ted Cruz and Trump Administration official Mercedes Schlapp.
a. Did these reported meetings take place? If so, what was the impetus for
scheduling these meetings? Were they requested by you or by the persons
with whom you met?
b. What topics were discussed at these meetings? What specific requests were
made of you at these meetings? Please provide any documents, meeting notes,
or memoranda relating to the deliberation.
c. To what extent was potential future regulation of social media platforms
generally or Twitter specifically discussed with current government officials?
Please provide any documents, meeting notes, or memoranda relating to the
deliberation.
d. To what extent have internal deliberations at Twitter regarding these
accusations of bias focused on potential future regulation of social media
platforms generally or Twitter specifically? Please provide any documents,
meeting notes, or memoranda relating to the deliberation.
e. Have any similar meetings with progressive or Democratic leaders taken
place, been offered, or been requested? Please provide any documents,
meeting notes, or memoranda relating to the deliberation.
f. Have any similar meetings with victims’ rights, anti-harassment, civil rights,
or anti-hate group advocates, or any other individuals or groups concerned
with rampant racist and misogynist harassment on Twitter taken place, been
offered, or been requested? Please provide any documents, meeting notes, or
memoranda relating to the deliberation.
72
g. Have any similar meetings with election security , campaign finance, or good
government advocates, or any other individuals or groups concerned with
Twitter’s role in the functioning of our democratic institutions taken place,
been offered, or been requested? Please provide any documents, meeting
notes, or memoranda relating to the deliberation.
Twitter officials have in the past and will continue in the future to meet with government
and political representatives across the political spectrum. We have good partnerships with civil
society groups around the globe, including with partners focused on election integrity. In every
interaction, we strive to uphold the values of the company.
Twitter does not use political ideology to make any decisions, whether related to ranking
content on our service or how we enforce our rules. We believe strongly in being impartial, and
we strive to enforce our rules impartially. We do not shadowban anyone based on political
ideology. In fact, from a simple business perspective and to serve the public conversation,
Twitter is incentivized to keep all voices on the platform.
Twitter plays an important role in our democracy and governments around the world. In
the United States, all 100 Senators, 50 governors, and nearly every member of the House of
Representatives currently reach their constituents through Twitter accounts. Our service has
enabled millions of people around the globe to engage in local, national, and global
conversations on a wide range of issues of civic importance. We also partner with news
organizations on a regular basis to live-stream congressional hearings and political events,
providing the public access to important developments in our democracy. The notion that we
would silence any political perspective is antithetical to our commitment to free expression.
2. After reported “anti-conservative” bias, the accusations of which are questionable, at
Facebook, specifically regarding its "Trending" section, Facebook removed human editors
from the process and used algorithms to moderate the Trending section.
Almost
immediately, the Trending section was overrun with false information and hoaxes
masquerading as news stories.
a. To what extent has Facebook’s experience with charges of anti-conservative
bias, its overcorrection, and its resulting issues with “fake news” informed
Twitter' s decision-making regarding similar issues?
b. What value do you, Twitter as a company, and Twitter's algorithms place on
veracity? How does Twitter balance competing concerns over equal
treatment of political viewpoints and legitimate concerns about
misinformation and harassment?
c. Is the deliberate spreading of misinformation a cause for suspending or
banning a Twitter account? Please provide any documents, meeting notes, or
memoranda relating to Twitter's formal policies on the subject.
73
d. Does Twitter have rules regarding the veracity of promoted content? If so,
what guidelines and processes are in place to evaluate the veracity of
promoted content? If not, have there been any internal discussions or
deliberations about moderating promoted content to prevent the spread of
misinformation?
Twitter is committed to help increase the collective health, openness, and civility of
public conversation, and to hold ourselves publicly accountable towards progress. At Twitter,
health refers to our overall efforts to reduce malicious activity on the service, including
malicious automation, spam, and fake accounts. Twitter has focused on measuring health by
evaluating how to encourage more healthy debate, and critical thinking.
The platform provides instant, public, global messaging and conversation, however, we
understand the real-world negative consequences that arise in certain circumstances. Twitter is
determined to find holistic and fair solutions. We acknowledge that abuse, harassment, troll
armies, manipulation through bots and human-coordination, misinformation campaigns, and
increasingly divisive echo chambers occur.
We have learned from situations where people have taken advantage of our service and
our past inability to address it fast enough. Historically, Twitter focused most of our efforts on
removing content against our rules. Today, we have a more comprehensive framework that will
help encourage more healthy debate, conversations, and critical thinking.
We believe an important component of improving the health on Twitter is to measure the
health of conversation that occurs on the platform. This is because in order to improve
something, one must be able to measure it. By measuring our contribution to the overall health of
the public conversation, we believe we can more holistically approach our impact on the world
for years to come.
Earlier this year, Twitter began collaborating with the non-profit research center Cortico
and the Massachusetts Institute of Technology Media Lab on exploring how to measure aspects
of the health of the public sphere. As a starting point, Cortico proposed an initial set of health
indicators for the United States (with the potential to expand to other nations), which are aligned
with four principles of a healthy public sphere. Those include:
● Shared Attention: Is there overlap in what we are talking about?
● Shared Reality: Are we using the same facts?
● Variety: Are we exposed to different opinions grounded in shared reality?
● Receptivity: Are we open, civil, and listening to different opinions?
74
Twitter strongly agrees that there must be a commitment to a rigorous and independently
vetted set of metrics to measure the health of public conversation on Twitter. And in order to
develop those health metrics for Twitter, we issued a request for proposal to outside experts for
their submissions on proposed health metrics, and methods for capturing, measuring, evaluating
and reporting on such metrics. Our expectation is that successful projects will produce
peer-reviewed, publicly available, open-access research articles and open source software
whenever possible.
As a result of our request for proposal, we are partnering with experts at the University of
Oxford and Leiden University and other academic institutions to better measure the health of
Twitter, focusing on informational echo chambers and unhealthy discourse on Twitter. This
collaboration will also enable us to study how exposure to a variety of perspectives and opinions
serves to reduce overall prejudice and discrimination. While looking at political discussions,
these projects do not focus on any particular ideological group and the outcomes will be
published in full in due course for further discussion.
In regards to advertising, Twitter has stricter rules and higher standards for promoted
content. Additional information on prohibited content policies can be found here:
https://business.twitter.com/en/help/ads-policies/prohibited-content-policies/inappropriate-conten
t.html
3. Relatedly, after Twitter took steps to remove “trolls” and fake accounts, conservatives
accused Twitter of executing a purge of conservative accounts. In reality, Twitter was
responding to concerns about widespread use of bots and fake accounts during the 2016
election.
a. What do you and Twitter as a company consider Twitter's responsibilities and
obligations to our democratic institutions in light of your platform's growing
importance to campaigning and governing?
b. To what extent are you and Twitter as a company concerned that allegations of
anti-conservative bias are an attempt to prevent Twitter from taking necessary steps
to prevent the abuse of the platform that occurred during the 2016 election?
Please provide responses to these questions and any relevant documents, meeting notes, or
memoranda relating to these matters.
Twitter is committed to improving the collective health, openness, and civility of public
conversation on our platform. Twitter’s is built and measured by how we help encourage more
healthy debate, conversations, and critical thinking. Conversely, abuse, malicious automation,
and manipulation detracts from it. We are committing Twitter to hold ourselves publicly
accountable towards progress.
The public conversation occurring on Twitter is never more important than during
elections, the cornerstone of our democracy. Our service shows the world what is happening,
75
democratizes access to information and—at its best—provides people insights into a diversity of
perspectives on critical issues; all in real-time. We work with commitment and passion to do
right by the people who use Twitter and the broader public. Any attempts to undermine the
integrity of our service is antithetical to our fundamental rights and undermines the core tenets of
freedom of expression, the value upon which our company is based. This issue affects all of us
and is one that we care deeply about as individuals, both inside and outside the company.
Additional information regarding our elections integrity efforts can be found here:
https://about.twitter.com/en_us/values/elections-integrity.html
76
@MattGaetz
6/4/2018
24,706 Followers
9/3/2018
Followers 46,291
@JimJordan
6/4/2018
163,620 Followers
9/3/2018
Followers 298,994
@MarkMeadows
6/4/2018
90,479 Followers
9/3/2018
156,621 Followers
@DevinNunes
6/4/2018
143,503 Followers
9/3/2018
256,216 Followers
