�鶹������ý

Google, one of the world’s largest technology companies, sees roughly 3.5 billion searches per day.¹ As the 2020 election approaches, people are likely going to increase their voting and election-related searches online, and it will be critical for Google to combat misleading information to ensure that its users can successfully participate in the electoral process.²

Google does not have content guidelines or restrictions for the websites that appear within its search engine. According to Google, its goal is to identify relevant information on the web based on users’ queries, not to decide whether the material is “in some sense wrongful.”³ However, due to recent concerns around election misinformation, Google announced on September 10, 2020 that it updated its search function to no longer suggest autocomplete search phrases for users looking up information on candidates or voting.⁴ Searches for informational claims on a candidate, how to cast a ballot, or the legitimacy of the electoral process will need to be typed out fully. The search engine will still suggest phrases for questions around voting or the election, i.e how to vote, but it will not autocomplete claims about voting or the election, i.e. you can not vote by mail. Google stated that it made this change to help prevent bad information from appearing in autocomplete suggested phrases.⁵

Apart from its search function, Google places requirements on advertising that runs on its platforms. According to Google’s ad policies, all ads must be “clear and honest, and provide the information that users need to make informed decisions.”⁶ The company prohibits ads that deceive users by including misleading information about products, services, or businesses. This includes the use of false claims or “deceptively doctoring media related to politics, social issues, or matters of public concern.”⁷ As of September 2020, Google’s misrepresentation policy for ads also prohibits accounts from coordinating with other sites or accounts to conceal or misrepresent their identities or other material details if the content of the account relates to politics, social issues, or matters of public concern.⁸ Although these general policies can help Google combat election and voting-related misinformation in advertising, the company does not have policies that specifically address ads aimed at suppressing voting. To ensure that these types of ads would not be allowed to run on its platform, Google’s policies should be updated to specifically address voter suppression content as well as election-related misinformation, especially leading up to the 2020 election.

Google states that accounts that violate its updated misrepresentation policy will be suspended “upon detection and without prior warning” and will not be allowed to advertise on Google’s platform in the future.⁹ It is unclear if Google’s standard appeals process for advertisers would apply to these accounts. Typically, if an advertiser feels that their ad was unfairly removed, they can appeal the decision directly in Google Ads, Google’s online advertising platform, by hovering over a disapproved or limited ad and clicking on an “appeal” link.¹⁰

Due to the COVID-19 pandemic, Google reduced its moderation and review workforce capacity because the number of people going into offices decreased and some content moderation tasks dealing with sensitive customer data or graphic content must be done onsite for security and wellness purposes.¹¹ The company warned that this may result in more content moderation mistakes as well as slower appeals review processes.¹² Given that a significant amount of election-related content moderation is occurring during the pandemic, Google should preserve data related to election-related content removals during this period so that researchers can evaluate these efforts later on. Additionally, the company should ensure that users and advertisers have access to a robust and timely appeals process. It is important for the company to remove misleading ads, but accurate voting and election-related ads on Google can be beneficial to encourage active participation and keep users informed, and advertisers who have their ads erroneously flagged should have the opportunity to appeal these decisions.

Google’s general ad targeting and delivery practices have raised concerns in the past for infringing on users’ privacy and for generating discriminatory outcomes.¹³ In an effort to increase accountability for political ads and help combat misleading information on the platform, Google updated its political ad policies in November 2019.¹⁴ Under Google’s updated policies, political ads must “comply with local legal requirements, including campaign and election laws for any geographic areas they target.”¹⁵ This covers any ad that is for a political organization, political party, political advocacy or fundraising event, or any ad pertaining to an individual candidate or politician.¹⁶

While Google claims that it has never allowed “granular microtargeting” for election ads, under the new policies, election-related campaigns have additional targeting limitations compared to non-election campaigns.¹⁷ For example, election ads may only target users based on geographic location (i.e. state or zip code), age, gender, or contextual targeting such as by topic or keywords against sites. Election ads may not utilize Google’s other advertising options such as remarketing, customer matches, geographic radius targeting, or third-party audiences. Limitations on microtargeting and other granular targeting, like radius targeting, are important because these techniques can be used to divide the public into very small groups,¹⁸ allowing misinformation or misleading content to avoid scrutiny from the broader public and spread more easily.¹⁹ While having a ban on geographic radius targeting can have a positive impact on ad visibility, Google does allow political advertisers to target their ads based on other geographic information, including zip code. Zip code targeting can be problematic when it comes to ads for housing, employment, and credit opportunities, because zip codes can be used as a proxy for race.²⁰ While zip code targeting could be beneficial for election related ads on local elections or for a presidential candidate attempting to increase their visibility in a certain region, it is worth examining whether advertisers have been able to use zip code targeting to promote voter suppression tactics for the presidential election.²¹

Along with targeting restrictions, Google requires identity verification for advertisers running federal or state election ads.²² There is a specific U.S. election ads verification form that requires an organization or individual to provide a Google Ads Customer ID and email.²³ Google reviews the application and provides an email notification within three to five business days. If the platform discovers that an account has violated election-related advertising policies or provided false information during the verification process, the account’s verification is revoked. Political ads are also required to have an in-ad disclosure that displays the name of the party or individual paying. For most ad formats, the “paid for by” disclosure is automatically generated from the account’s information. Otherwise, it is the advertiser’s responsibility to add the disclosure.²⁴

Google publishes a transparency report to keep the public informed about political advertising on Google, YouTube, and partner properties.²⁵ The report includes information such as the number and the amount of money spent on political ads since May 31, 2018, top advertisers by total ad spend, and ad spend by state. It also allows people to view or search for ads. Although these data points are a good starting point, the report does not break down certain information, such as ad type (i.e. search versus display), or provide granular details such as the type or amount of policy-violating ads that have been removed. In addition, the report only includes ads that feature a “current officeholder or candidate for an elected federal or state office, federal or state political party, or state ballot measure, initiative, or proposition that qualifies for the ballot in a state.” As a result, the report does not provide a comprehensive overview of all political ads that are run on the platform, and it creates a significant gap in terms of which ads are available for public scrutiny.²⁶ In addition, Google does not currently publish a comprehensive transparency report outlining its content policy enforcement for its web search product (it does publish one for YouTube).²⁷ Going forward, the company should expand its transparency reporting efforts to include information on policy-violating content, and ensure that the report includes data on the enforcement of its election-related misinformation and disinformation policies. Further, in the run up to 2020 elections, the company should provide periodic updates on the enforcement of its election-related misinformation content policies and ad policies, including how much user content and how many accounts have been removed, and how many ads have been rejected and removed for violating election-specific policies or supporting voter suppression. Following the 2020 presidential election, the company should issue a comprehensive transparency report detailing its moderation efforts that includes data on voter suppression and election related content.

Outside of advertising, Google also offers Google News, an aggregator that helps connect users to different news and media outlets.²⁸ The platform utilizes an algorithm to determine how news is ranked, which relies on factors such as relevance of content, prominence, location, and language.²⁹ Google does not accept payments to expedite or increase a site’s visibility within the news aggregator.³⁰ To appear in Google News, a publisher does not need to submit their site for approval, they simply need to “produce high-quality content and comply with Google News content policies.”³¹ However, it is unclear how Google decides what type of content qualifies as high-quality. Under Google News’ community policies for publishing partners, content may not contain deceptive practices or attempt to mislead users.³² The platform also does not allow sites or accounts to impersonate any person or organization or misrepresent their primary purpose. However, Google News’ policies do not specifically address election or voter suppression content.³³ Apart from providing information on how its algorithm ranks news sources, Google News should clarify how it first selects which news sources to include. The platform should also clarify if it takes steps to monitor the news sources for potential voter suppression and election misinformation content.

According to Google, in order to combat the spread of misinformation on its platforms, its Trust and Safety and Threat Analysis Group (TAG) teams monitor and combat inauthentic activity, disinformation campaigns, and other forms of abuse across Google’s platforms on a 24/7 basis.³⁴ Google utilizes Jigsaw, a unit within the company that forecasts and fights against emerging threats, to help combat harassment and misinformation across a variety of topics, including voting and election information.³⁵ Google explains that Jigsaw’s technology enables “deepfake” detection, therefore allowing the company to protect user accounts and campaigns that are targets of hacks or phishing attacks.³⁶ Additionally, Google works with other technology companies and government agencies, such as the FBI’s Foreign Influence Task Force, to help combat harmful interference online.³⁷

While Google has taken steps to combat voter and election related misinformation, the measures have not always proven effective. According to a Tech Transparency Project (TTP) analysis, Google failed to remove ads linking to websites that charged fees, in some cases fees up to $129, to register to vote, although eligible voters can typically register directly with their states or territories for free.³⁸ From the analysis, it is unclear whether Google failed to remove these ads because they were overlooked or because they did not explicitly violate Google’s policies. These types of ads can suppress voting because people may not be able to afford the supposed registration fees, or may not be willing to pay to register to vote. Also, if online users looking for information about the election are sent to confusing websites or feel they are being manipulated, they may give up on trying to find important voting-related information. Furthermore, users may be directed to websites that are trying to extract personal information or receive fraudulent donations, which can have negative repercussions beyond voter suppression.

On August 13, Google released a blog post with additional steps they are taking to combat misinformation around the 2020 U.S. election.³⁹ The platform released two new features in Google Search that allow users to search for detailed information on how to register and when to vote by state, by searching “how to vote.” Google collects this information from Democracy Works, a third-party non-partisan organization that collects data directly from state and county election administrators.⁴⁰ The post also states that Google is reaffirming its investment in the company’s TAG and Trust and Safety teams, which are assigned to work together to prevent government-backed attacks and phishing campaigns on Google users. TAG tracks over 270 targeted or government-backed groups from over 50 countries, whose goals include “intelligence collection, stealing intellectual property, targeting dissidents and activists, destructive cyber attacks, or spreading coordinated disinformation.”⁴¹ Google, along with other major tech companies including Facebook, Pinterest, Reddit, and Twitter, also participates in meetings with government agencies that are responsible for election integrity to discuss trends in the spread of misinformation.

While Google has been vocal about taking measures to combat voter and election-related misinformation and disinformation, the company should go further in providing transparency around how effective these efforts have been. For example, the company should include statistics in its transparency report on how many ads have been removed for violating ad policies and on how many government-backed attacks it stopped leading up to the 2020 election.