Written Testimony of Derek Slater  Director Information Policy Google LLC    United States Senate Committee on Commerce Science and Transportation  “Mass Violence Extremism and Digital Responsibility”  September 18 2019    Chairman Wicker Ranking Member Cantwell and distinguished members of the  Committee Thank you for the opportunity to appear before you today I appreciate  Congress’ work in looking closely at how to prevent tragic episodes of mass violence      My name is Derek Slater and I am the Global Director of Information Policy at Google   In that capacity I lead a team that advises the company on public policy frameworks  for dealing with online content -- including hate speech extremism and terrorism   Prior to my role at Google I worked on Internet policy at the Electronic Frontier  Foundation and at the Berkman Center for Internet and Society     Before I begin I would like to take a moment on behalf of everyone at Google to  express our horror in learning of the tragic attacks in Texas and Ohio and to share our  sincere condolences to the affected families friends and communities While Google  services were not involved in these recent incidents we have engaged with the White  House Congress and governments around the globe on steps we are taking to  ensure that our platforms are not used to support hate speech or incite violence         We believe the free flow of information and ideas has important social cultural and  economic benefits though society has always recognized that free speech must be  subject to reasonable limits This is true both online and off and it is why in addition to  respecting the law we have additional policies procedures and community guidelines  that govern what activity is permissible on our platforms      In my testimony today I will focus on three key areas where we are making progress  to help protect people i how we work with governments and law enforcement ii   our efforts to prohibit the promotion of products that cause damage harm or injury   and iii the enforcement of our policies around terrorism and hate speech     Working with Government and Law Enforcement    Google appreciates that law enforcement agencies face significant challenges in  protecting the public against crime and terrorism Google engages in ongoing  dialogue with law enforcement agencies to understand the threat landscape and  respond to threats that affect the safety of our users and the broader public When  we become aware of statements on our platform that constitute a threat to life or that  reflect that someone’s life may be in danger we report this activity to law  enforcement agencies      For example when we have a good faith belief that there is a threat to life or serious  bodily harm made on our platform in the United States the Google CyberCrime  Investigation Group CCIG will report it to the Northern California Regional  Intelligence Center NCRIC In turn NCRIC quickly gets the report into the hands of  officers to respond CCIG is on call 24 7 to make these reports      Under U S law the Stored Communications Act allows Google and other service  providers to voluntarily disclose user data to governmental entities in emergency  2    circumstances where the provider has a good faith belief that disclosing the  information will prevent loss of life or serious physical injury to a person Our team is  staffed on a 24 7 365 basis to respond to these emergency disclosure requests  EDRs We have seen significant growth in the volume of EDRs that we receive from  US governmental entities as illustrated in our t​ ransparency report covering  government requests for user data​ In fact the number of EDRs submitted from  agencies in the US almost doubled from 2017 to 2018 We have grown our teams to  accommodate this growing volume and to ensure we can quickly respond to  emergency situations that implicate public safety     We are also deeply committed to working with government the tech industry and  experts from civil society and academia to protect our services from being exploited  by bad actors The recent tragic events in Christchurch presented unique challenges   and we had to take unprecedented steps to address the sheer volume of new videos  related to the events In the months since Google and YouTube signed the  Christchurch Call to Action a series of commitments to quickly and responsibly  address terrorist content online This is an extension of our ongoing commitment to  working with our colleagues in the industry to address the challenges of terrorism  online Since 2017 we’ve done this through the Global Internet Forum to Counter  Terrorism GIFCT of which Google is a founding company and was its first chair   Recently GIFCT introduced joint content incident protocols for responding to  emerging or active events The GIFCT also released its first-ever Transparency Report  and a new counterspeech campaign toolkit that will help activists and civil society  organizations challenge the voices of extremism online     Prohibiting the Promotion of Products That May Cause Damage Harm or Injury    We take the threat posed by gun violence in the United States very seriously and our  advertising policies have long prohibited the promotion of weapons ammunition   3    explosive materials fireworks and similar products that cause damage harm or  injury Similarly we also prohibit the promotion of instructions for making guns   explosives or other harmful products     On platforms like Google Ads and Google Shopping Ads we employ a number of  proactive and reactive measures to ensure that our policies are appropriately  enforced For example we run automated and manual checks to detect content that  violates our policies If an advertiser or merchant violates our policies we will take  appropriate action up to and including suspension of their account Users can also  provide direct feedback on ads that potentially violate Google policies via an external  form using the ‘Report a violation’ link or via the​ ​feedback link on Google com and  other Google properties to report any products that may violate our policies This  feedback is reviewed by our teams and appropriate action is taken      We know that we must be vigilant on these issues and are constantly improving our  enforcement procedures including implementing enhancements to our automated  systems and updating our incident management and manual review procedures     Policies and Enforcement on YouTube for Terrorism and Hate Speech    We have robust policies and programs to defend our platforms to spread hate or  incite violence This includes prohibitions on terrorist recruitment violent extremism   incitement to violence glorification of violence and instructional videos related to  acts of violence We apply these policies to violent extremism of all kinds whether  inciting violence on the basis of race or religion or as part of an organized terrorist  group     4    In order to improve the effectiveness of our policy enforcement we have invested  heavily in both technology and people to quickly identify and remove content that  violates our policies against incitement to violence and hate speech     1 YouTube’s enforcement system starts from the point at which a user uploads a  video If our technology detects that the video is similar to videos that we know  already violate our policies it is sent for humans to review If they determine  that it violates our policies they remove it and the system makes a “digital  fingerprint” or hash of the video so it can’t be uploaded again      2 Machine learning technology also helps us more effectively identify this  content and enforce our policies at scale However because hate and violent  extremism content is constantly evolving and can sometimes be  context-dependent we also rely on experts to help us identify policy-violating  videos Some of these experts sit at our intel desk which proactively looks for  new trends in content that might violate our policies We also developed an  improved escalation pathway for expert NGOs and governments to notify us of  bad content in bulk through our Trusted Flagger program We reserve the final  decision on whether to remove videos they flag but we benefit immensely  from their expertise     3 This broad cross-sectional work has led to tangible results Over 87% of the 9  million videos we removed in the second quarter of 2019 were first flagged by  our automated systems More than 80% of those auto-flagged videos were  removed before they received a single view And overall videos that violate our  policies generate a fraction of a percent of the views on YouTube     Our efforts do not end there as we are constantly evolving to new challenges and  looking for ways to improve our policies For example YouTube recently updated its  5    Hate Speech policy to specifically prohibit videos alleging that a group is superior in  order to justify discrimination segregation or exclusion based on qualities like age   gender race caste religion sexual orientation or veteran status This would include   for example videos that promote or glorify Nazi ideology because it is inherently  discriminatory YouTube also updated its policies to prohibit content denying that  well-documented violent events like the Holocaust or the shooting at Sandy Hook  Elementary took place      The updated Hate Speech policy was launched in early June and as our teams review  and remove more content in line with the new policy our machine learning algorithms  will improve in tandem to help us identify and remove such content Though it can  take months for us to ramp up enforcement of a new policy the profound impact of  our Hate Speech policy update is already evident in the data released in this quarter’s  Community Guidelines Enforcement Report​ the number of individual video removals  for hate speech saw a 5x spike to over 100 000 the number of channel terminations  for hate speech also saw a 5x spike to 17 000 and the total comment removals nearly  doubled in Q2 to over 500 million due in part to a large increase in hate speech  removals      Finally we go beyond removing policy-violating content by actively creating programs  to promote beneficial counterspeech These programs present narratives and elevate  credible voices speaking out against hate violence and terrorism For example our  Creators for Change program supports creators who are tackling tough issues   including extremism and hate by building empathy and acting as positive role models   We launched our most recent Creators for Change global campaign videos in  November 2018 As of June 2019 they already had 59 million views the creators  involved have over 60 million subscribers and more than 8 5 billion lifetime views of  their channels and through ‘Local Chapters’ of Creators for Change creators tackle  challenges specific to different markets   6      Alphabet’s Jigsaw group an incubator to tackle some of the toughest global security  challenges has deployed the Redirect Method which uses targeting tools and curated  YouTube playlists to disrupt online radicalization The method is open to anyone to  use and NGOs have sponsored campaigns against a wide-spectrum of  ideologically-motivated terrorists and violent extremists      Conclusion    We take the safety of our users very seriously and value our close and collaborative  relationships with law enforcement and government agencies We have invested  substantial resources to tackle the problem of hate speech At present we spend  hundreds of millions of dollars annually and have more than 10 000 people working  across Google to address content that might violate our policies which include our  policies against promoting violence and terrorism     We understand these are difficult issues of great interest to Congress and want to be  responsible actors who are a part of the solution As these issues evolve Google will  continue to invest in the people and technology to meet the challenge We look  forward to continued collaboration with the Committee as it examines these issues   Thank you for your time I look forward to taking your questions   7