We were able to act quickly because of the investments we made to prepare for the 2020 elections. Last year, teams across Google and YouTube worked around the clock to contribute to election preparedness, by helping voters find authoritative information about the election; by working with campaigns to equip them with best-in-class security features and helping them connect with voters; and by protecting our platforms from abuse.
Helping voters find authoritative information on our services
This U.S. election cycle saw all-time highs in searches on Google for civics-related topics. Anticipating that need, we worked to launch features that would help people find the information to participate in the democratic process, including how to register and how to vote in their states.
Consistent with our approach to prior election cycles, we showed “how to register” and “how to vote” reminders to all our U.S. users directly on Google Search, Maps and YouTube. These reminders were seen over 2 billion times across our products. As the election neared, we helped people find polling and ballot drop off locations: From mid-October through Election Day, we added more than 125,000 voting locations in Google Maps. Across our products, these features were seen nearly 500 million times.
Finally, starting on Election Day, we worked with the Associated Press to provide real-time election results for relevant searches on Google. These results had over six times more views in 2020 than in 2016. Similarly, on YouTube, we launched an election results information panel that showed on top of search results and under videos with election-related content. It pointed to our election results page on Google, and over time, we expanded it to include an additional link pointing to a page on the U.S. Cybersecurity and Infrastructure Security Agency (CISA) website that debunked incorrect claims made about the integrity of the elections. Once the safe harbor deadline for state certification passed, we updated this YouTube Election Results Information Panel again to point to the National Archives Office of the Federal Register page of record for the 2020 electoral college vote. Collectively, our election information panels on YouTube have been shown over 8 billion times.
Working with campaigns
We also helped campaigns and elected officials effectively use Google and YouTube products to reach voters and enhance their election security. As part of our Civics Outreach Virtual Training Series, Google held 21 training sessions for over 900 candidates, campaigns, public officials, and nonprofit leaders. Overall, we held 45 group and individual trainings to help more than 2,900 election workers learn to use Google tools to amplify their message and better connect with voters through events like digital town halls, debates and virtual campaign rallies.
In addition, as a part of our Election Cybersecurity Initiative with the University of Southern California’s Annenberg School, nearly 4,000 elected officials, secretaries of state, campaign staffers, political party representatives, and state election directors in all 50 states received training on ways to secure their information and protect their campaigns against cyberattacks.
At the start of the 2020 election season, we partnered with Defending Digital Campaigns (DDC), a nonprofit and nonpartisan organization, to give any eligible federal campaign access to free Titan Security Keys — the strongest form of two-factor authentication. This collaboration is a part of our Advanced Protection Program, which protects high-risk individuals, such as election officials, campaigns, and journalists, who have access to high visibility and sensitive information. In the lead-up to the 2020 elections, DDC distributed more than 10,000 Titan Security key bundles to more than 140 U.S. federal campaigns. We recently expanded our support for DDC to provide eligible campaigns and political parties, committees, and related organizations, at both the federal and state levels, with knowledge, training and resources to defend themselves from security threats.
Protecting our platforms from abuse
In the years leading up to the 2020 election, we made numerous enhancements to protect the integrity of elections around the world and better secure our platforms. Among them, we introduced strict policies and processes for identity verification for advertisers who run election-related advertising on our platform; we launched comprehensive political ad libraries in the U.S. and in other countries around the world; we developed and implemented policies to prohibit election-related abuse such as voter suppression and deceptive practices on platforms like YouTube, Google Ads, Google Maps and Google Play; our Threat Analysis Group (TAG) launched a quarterly bulletin to provide regular updates on our work to combat coordinated influence operations across our platforms and flagged phishing attempts against the presidential campaigns; and we worked closely with government agencies, including the FBI’s Foreign Influence Task Force, and other companies to share information around suspected election interference campaigns.
On YouTube, throughout 2020, we identified and removed content that was misleading voters about where or how to vote, to help ensure viewers saw accurate information about the upcoming election. After December 8, which marked the “safe harbor” deadline for states to certify their election, in accordance with our Presidential Election Integrity policy we began to remove content uploaded on or after December 9 that misled people by alleging that widespread fraud or errors changed the outcome of the 2020 U.S. presidential election. In addition, we continued to enforce our broader policies — for instance, from October to December 2020, we removed 13,000 YouTube channels for promoting violence and violent extremism; 89% of videos removed for violating our violent extremism policy were taken down before they had 10 views.
This work was in addition to improvements in the ranking systems we use to reduce the spread of harmful misinformation on YouTube: In January 2019, we announced that we would begin reducing recommendations of borderline content or videos that could misinform viewers in harmful ways but that do not violate YouTube Community Guidelines. Since then, we’ve launched numerous changes to reduce recommendations of borderline content and harmful misinformation, and we continue to invest in this work: Our models review more than 100,000 hours of videos every day to find and limit the spread of borderline content.
Our work is never done, and we continue to learn and improve from one election cycle to the next, and continue to evolve our policies. That principle has guided our approach to new and evolving challenges, including COVID-19 misinformation.