Testimony of Maya Wiley Before the U.S. Senate Committee on Rules and Administration on AI and the Future of Our Elections

Uncategorized 09.27,23

View a PDF of the testimony here.

STATEMENT OF MAYA WILEY, PRESIDENT AND CEO
THE LEADERSHIP CONFERENCE ON CIVIL AND HUMAN RIGHTS

UNITED STATES SENATE
COMMITTEE ON RULES AND ADMINISTRATION

HEARING ON “AI AND THE FUTURE OF OUR ELECTIONS”

September 27, 2023

I. Introduction

Chair Klobuchar, Ranking Member Fischer, and members of the committee: Thank you for the opportunity to testify today. My name is Maya Wiley, and I am the president and CEO of The Leadership Conference on Civil and Human Rights, the nation’s oldest and largest civil rights coalition with a diverse membership of more than 240 national organizations working to build an America as good as its ideals. Our coalition is dedicated to democracy, which requires promoting and protecting the civil and human rights of every person in the United States. Our work is nonpartisan, from the fight for voting rights and every bipartisan reauthorization of the Voting Rights Act of 1965 to the continued efforts to ensure voters have access to the ballot and the ability to cast their ballots without interference.

To support fair and non-discriminatory elections, we also work to ensure voters know how to register, where to cast their ballots, and how engage in the civic process as poll workers. With our members and partners, we monitor election cycles for misinformation and disinformation to ensure that voters are not misled, deceived, or threatened by others. We actively engage social media platforms in the types of policies and practices that prevent and reduce hate, bias, and deception online to ensure that our democracy works for all of us.

We have been working at the intersection of civil rights and technology for more than a decade, and we have long been leading the fight against online hate, bias, and mis- and disinformation. We recently announced the establishment of a first-of-its-kind Center for Civil Rights and Technology. This Center will expand and deepen our long-standing work on media and technology, and it will actively work with our coalition, advocates, academics, policymakers, and the private sector on the positive possibilities of artificial intelligence (AI). This work will help to ensure a regulatory structure, transparency, and accountability along with other guardrails necessary to ensure that AI supports a country where all people and communities reap the benefits of, rather than suffer harm from, rapidly transforming technology.

Today, I will share two observations from our coalitional work on free and fair elections: First, technology is a tool, but we must have checks on its intentional misuse, consistent with Constitutional protections; and second, we must ensure fair elections, free from bias and discrimination, which reinforces what we know to be true: We are fortunate to have our democracy.

Sixty years ago, in August of 1963, more than 200,000 people gathered in Washington, D.C., for the March on Washington for Jobs and Freedom. The Leadership Conference was there with our founders. From the steps of the Lincoln Memorial, one of our founders — union leader A. Philip Randolph, president of the Brotherhood of Sleeping Car Porters — said that “The March on Washington is not the climax of our struggle, but a new beginning not only for the Negro but for all Americans who thirst for a better life.” This remains true today. And now, it is technology that poses a potential opportunity, but also a rapidly evolving set of risks. This includes the poisonous bloom of disinformation and hate that can threaten free and fair elections.

Our democracy depends on trust — trust in the integrity of our elections, trust in the information we receive about our elections, and trust that when we participate, we can do so safely. Democracy also depends on trust in our leaders — trust that they will be truthful about the facts, even when rational minds can differ about how to interpret those facts. Mis- and disinformation are clear threats, and their spread has already threatened the peaceful transfer of power after our 2020 presidential election.

In 2018, many of us watched a YouTube video, purportedly of Barack Obama, discussing the ability of an “enemy” using technology to make it look like someone was saying something they never said.[1] About halfway through the video, the screen splits and we see that the voice is not from the former president but from the actor, comedian, and filmmaker Jordan Peele imitating Barack Obama. This educational video gave us a glimpse of the future of mis- and disinformation in the form of deepfakes and their ability to deceive our people and undermine our national security. Artificial intelligence is not new, but it is rapidly developing thanks to breakthroughs in generative technology. In the 2020 election, we witnessed little in the way of deepfakes when compared to today — but it was a risk even then.

Even before the rapid advances we are seeing in AI and the tools they enable for the rapid spread of disinformation, we had disinformation intentionally targeting communities of color.[2] As the Associated Press reported, in 2020, Facebook ads targeting Latino and Asian communities in Chicago falsely claimed candidate Joe Biden was a communist. A doctored photo was altered from the true image of a dog urinating on a Biden campaign poster to a fake image of the dog urinating on a Trump campaign poster.[3] The Senate Intelligence Committee found that Russia-backed operatives aggressively targeted disinformation toward African Americans during the 2016 election to suppress votes, stating, “[N]o single group of Americans was targeted by IRA information operatives more than African Americans.” We have seen enough to know we must be prepared for what is upon us.[4]

II. The Origins of Artificial Intelligence on Our Democracy

During the 2020 and 2022 election cycles, The Leadership Conference and our partner Common Cause monitored, analyzed, and responded to mis- and disinformation. Most of the content that we flagged falls into the narratives and trends that we have seen over the last two years, including the Big Lie — the phrase used to describe false claims that Joe Biden did not win the 2020 election — and false information about elections processes, such as mail-in ballots and the use of ballot drop boxes. We used the flagged content as evidence to inform the platforms and the government of election disinformation trends/issues in private conversations and in public letters and statements. What we have found is also supported by academic research, including:

  • Online mis- and disinformation continues to confuse, intimidate, and harass people, including voters; suppress the right to vote; spread hate speech; and otherwise disrupt our democracy.
  • Vulnerable communities and communities of color are disproportionately threatened. They are often the intended recipients of mis- and disinformation intended to suppress the vote and drive wedges betwteen communities of color. They are also impacted online and offline by mis- and disinformation about their communities, including content that stirs up hate and distrust of these communities within other demographics.
  • One example of the ramifications of online disinformation includes:
    • After an election official in Colorado (who was an outspoken election denier) allegedly assisted in the posting of a video showing passwords used to access the county’s voting system that was secretly recorded during a security update, election workers statewide started receiving threats and resorted to wearing bulletproof vests and undergoing active shooter preparedness training in response to the possibility of right-wing violence.[5]

Fast-evolving technologies literally enable users to put words and actions into the mouths and bodies of real people that they never uttered or made.[6] Since 2017,[7] the use of “deepfakes” have become all too common, where the faces of people, largely women,[8] have been attached to the bodies of others in pornographic material, destroying reputations and even being used for blackmail and human trafficking.[9] While “cheapfakes,” such as photoshopping images and other forms of visual manipulation, are not new, AI is advancing the opportunities for intentional manipulation. Data and Society, an independent nonprofit research institute, recently stated in a report on deepfakes: “With thousands of images of many of us online, in the cloud, and on our devices, anyone with a public social media profile is fair game to be faked.”[10]

Deepfakes are already being used in national, state, and local races by supporters of candidates and political action committees, according to some researchers.[11] Just this past summer, Never Back Down, a group that supports Florida Governor Ron DeSantis’ candidacy for president, ran an ad in Iowa with a voice, generated by AI, which sounded like Donald Trump’s voice saying words he wrote on social media but never spoke.[12]

This is just one example, and there are many other national and local examples that have been widely reported, including:

  • In a Chicago mayoral race, an X account called “Chicago Lakefront News” posted a video that appeared to show candidate Paul Vallas saying, “in my day,” a police officer could kill as many as 17-18 civilians and “no one would bat an eye.” This video was entirely AI-generated, although it appeared authentic.[13]
  • A deepfake also circulated depicting Sen. Elizabeth Warren (D. Mass.) insisting that Republicans should be barred from voting in 2024.[14]
  • Earlier this year, an AI-generated video showing President Biden declaring a national draft to aid Ukraine’s war effort — initially acknowledged as a deepfake but later stripped of that context — led to a misleading tweet that garnered more than 8 million views.[15]

The rapid growth of AI since the 2022 elections has the potential to significantly increase the volume and speed of disinformation in the 2024 election cycle. The false narratives we have seen in prior elections are certain to come up again in 2024 in new forms with even more intensity. AI has the potential to turbocharge the volume and the spread of disinformation. Generative AI can increase the ability and scale to spread false information and propaganda, leaving voters confused and further questioning what they see and hear. Our coalition member, The Brennan Center, has pointed out that elections are particularly vulnerable to AI disinformation. Generative AI tools are most effective when they produce content similar to their current databases. Since the same false election narratives will likely come up again in future elections, there is an abundance of past election disinformation in the training data underlying generative AI tools that can make them a time bomb for future election disinformation.[16] The proliferation of AI-generated content could accelerate the loss of trust in the integrity and security of our overall election system and dramatically interfere with the right to vote.[17]

AI can spread disinformation, particularly toward communities of color, in a number of ways, including deepfaked audio and visuals. Chatboxes (e.g., ChatGPT) can spread false narratives further. AI could send false or deceptive comments from fake constituents or advocates or set up fake news sites. Chatboxes and deepfakes could threaten election systems through targeting and phishing efforts that are more targeted and personalized to users, making them potentially more effective.[18] Disinformation spreaders can plant false information on websites that generative AI can pick up and spread. While a benefit of AI is its support for non-English speakers to have translation tools, non-English speakers have been targeted by disinformation in elections.

Unfounded conspiracy theories are causing some states to withdraw from the Election Registration Information Center despite its bipartisan origins and the bipartisan praise it has received for supporting election integrity, including the ability to better identify deceased voters.[19]

Arizona, for example, attempted to provide more access to voters’ personal information, making targeting with powerful AI more effective and raising the threat that information will be abused through AI targeting.

This is just one of many examples of where comprehensive reform and regulation could allow for effective AI use, but underscoring a need for regulation so that use is not abused.

III. Eroding Voting Rights Creates Conditions for Disinformation to Thrive

Since our independence from England, the nation has spent more than 200 years, through protest and too often in the face of threats and violence, to make our democracy real for all citizens — from white men who did not have the wealth to own land until the 1820s, to the Suffragette movement and Native American enfranchisement in 1920, to Bloody Sunday and the Voting Rights Act of 1965 that, after 100 years of blatantly undermining the Black vote, finally provided meaningful protections for lawful voting without discrimination. The attacks on those hard-won protections have created the conditions in which mis- and disinformation are more easily cooked up and consumed by the American public.

Critical to those protections was the bipartisan Voting Rights Act and all its reauthorizations from 1965 through 2006, which included the clear support of four Republican presidents. The Voting Rights Act built trust in our democracy. After successive demands for a more inclusive democracy, it was not until the second half of the 20th century that we finally achieved the right to democratic inclusion, which did not come easily. When we did achieve democratic inclusion, it was bipartisan. Until 2010, we had roughly four decades of a relatively trustworthy and trusted democratic process. It has never been perfect, but the integrity of our voting systems, the trust we shared across our beliefs and our diversity, and the fact that — for the most part — ours is a trustworthy system, has been the envy of much of the world.

That began to change in 2010.[20] The historic turnout of voters, including Black voters and young voters of all races in the 2008 presidential race, highlighted the significant power that coalitions of voters brought to the polls and our democracy. Backed by deep pockets including many corporations, state legislation was introduced across the country that would create barriers to the ballot box for lawful voters.[21] Voter identification laws were becoming the most prevalent form of barriers to the ballot. The alleged justification for these restrictive laws was voter fraud,[22] often wrapped in fear mongering about undocumented immigrants. In some instances, party operatives openly stated that the laws were designed to make it harder for voters who were likely to vote for the opposing party.[23] Black voters in particular were targeted.[24] The claims were not only unsupported, but research over the past two decades makes clear that voter fraud has not been a significant problem.[25]

As noted in a recent report from The Leadership Conference Education Fund, in the 10 years since the U.S. Supreme Court’s Shelby County v. Holder ruling, state and local jurisdictions have used the absence of federal voting protections to try to take us backward by creating barriers to the ballot for Black, Brown, and Native voters; people with disabilities; young and older people; and new Americans.[26] This has created more fuel to fire mis- and disinformation. Well-funded state-by-state strategies — erected to create unnecessary barriers to the ballot — have created a tidal erosion of trust in the integrity of our elections and have done so by driving divisions and stoking fear in some instances that Mexican immigrants in particular — and Latino voters in general — were voting unlawfully. One of the central legal protections used to protect voters from partisan power grabs was the Voting Rights Act of 1965 and all its bipartisan reauthorizations. Years after the 2008 election, fake news about millions of undocumented immigrants voting in 2008 appeared on a Fox and Friends tweet.[27]

These unfounded statements supported attacks on vote by mail and stricter voter identification laws, voter roll purges, and the creation of voting police — intimidating marginalized voters or making it less likely they would vote. Older Americans, students, and people with disabilities saw improvement in their ability to have access to voting with mail-in ballots, early voting, and more polling sites, which were crucial for their ability to participate in our democracy.

The frontal and explicit attacks on the free and fair election of 2020 — and the false claims that President Biden is not a duly elected leader of this free republic — have resulted in unprecedented violence, and they have only served to deepen divisions and stoke everything from threats to harassment to outright violence against public servants and ordinary people, as well as elected leaders.[28]

State laws that suppress the vote and fuel distrust and division have continued to grow, resulting in an astounding number of state voter restrictions in recent years. State legislatures in 20 states enacted at least 33 new restrictions in effect for the 2022 midterms.[29]And already this year, at least 11 states have enacted 13 more restrictive voting laws.[30] As in 2010, this push to restrict voting access gained new intensity after voters of color made their voices heard through robust turnout in 2020. In a dangerous affront to basic concepts of a constitutional democracy, 26 states created or expanded existing criminal penalties related to voting. In total, 60 new felonies and 50 new misdemeanors were enacted after the 2020 election, including criminal penalties for assisting voters in some instances.[31] It amounts to state-driven voter intimidation, the likes of which we have not seen since the crush of state violence to prevent newly recognized Black citizens from voting and taking up public office after the end of the Civil War.

IV. Social Media Platforms Must Do More

Foreign governments, crime rings, organizations, and malicious individuals have actively created and disseminated baseless conspiracy theories about the integrity of our elections. They have spread dangerous, debunked, and factually unsupported claims about one of the deadliest pandemics in our history. This toxic content has caused immense harm — including, in some instances, acts of violence.

Social media platforms have not done their part. Unfounded claims of election fraud spread by social media platforms, according to researchers at the Stern School of Business at New York University,[32] amplified false claims about election denial, including on Meta, YouTube, X, and TikTok. Despite violations of their policies, these platforms either failed to take sufficient action to protect users of their platforms through potential corrective actions — such as increased labeling, demotion of content, or limiting the sharing of content — or allowed proponents of disinformation to exploit long-form videos.

As we have seen, this disinformation propelled the horrific acts of violence on our Capitol and our constitutional process for certifying Electoral College votes on January 6, 2021. But it didn’t stop there. Both organized extremists and distrusting and dangerous individuals have been empowered and incited by unfounded claims that some elected officials, business leaders, and celebrities have elevated and endorsed, which has deepened distrust and stoked the spread of hate, harassment, and harm — both in and between election cycles.[33]

There have been online and real-life attacks on election officials and poll workers, which threaten the operations and functioning of our election systems and intimidate the very people to whom we should extend our gratitude and our protection.[34] Disinformation from the “Big Lie” about voter fraud and intentional efforts to mislead voters on where and how to vote continue as drivers of threats and undermine our democratic practice of lawful voting. False information about the “Big Lie” is still spreading rapidly on social media, and it is the basis for the forthcoming spread of disinformation about the 2024 elections by election deniers and high-profile users.

The Leadership Conference and our coalition have repeatedly urged social media platforms to take immediate steps to curb the spread of voting disinformation and hate speech in the midterms and future elections to protect the health of our democracy. In May and October of 2022, our coalition sent letters that called on major platforms to take several affirmative steps well in advance of the midterm elections to combat election disinformation.[35] These include consistently enforcing existing policies, addressing election disinformation continuously, and taking action against non-English disinformation.

We also recently urged YouTube to reverse its recent decision to allow false election claims about the outcome of the 2020 election on its platform and reinstate the policy that prohibits 2020 election denial content on the platform.[36]

However, platforms did not and have not taken meaningful steps on the actions we have requested. We have seen:

  • Lack of enforcement of current voting disinformation policies
  • Reduction or elimination of trust and safety teams charged with monitoring content
    • YouTube shed two of its five policy experts who worked on hate speech and harassment issues, leaving only one person in charge of misinformation worldwide.[37]
    • According to an interview with the BBC in April, Elon Musk claimed that X had only about 1,500 employees, about an 80 percent drop from prior to his takeover.[38]
    • X completely eliminated the team that oversaw disinformation and trust and safety issues.[39] Sadly, X is not alone. Meta has cut 21,000 jobs over the last nine months, including deep cuts to trust and safety teams.[40]

Most major platforms’ enforcement of their own disinformation policies continues to be erratic and inconsistent, particularly against high-profile users, leading to false claims about voting processes and attempts to overturn the certified results of our elections. Platforms are actually loosening restrictions, such as YouTube now allowing content about the Big Lie to be posted at a time when platforms should be stepping up to detect and address AI-generated disinformation about voting.[41]

Despite the policies these platforms claim to have, we have tracked a systematic failure to enforce them. Despite these platforms’ immense profits and user bases — nearly 3 billion users for Facebook, more than 2.5 billion users for YouTube, and nearly 1.5 billion users for Instagram — public reporting shows that the platforms are cutting staff whose job is to help protect the public from dangerous content. A major social media platform simply cannot responsibly apply its policies with only one person responsible for political misinformation and two for medical misinformation.

The threat of election disinformation is still prevalent and far from over. There were disinformation narratives that took hold in 2022, particularly in contested states and where the counting of results took several days. We particularly saw this in Arizona and Nevada: While there was progress in addressing voting disinformation in 2022, election denialism will likely continue to be a persistent fixture in future campaigns. Disinformation around certification, counting results, and harassment of election officials will likely continue.

V. Potential Steps

The Leadership Conference and our coalition have been actively advocating for Congress and the administration to take action to ensure sufficient safety, transparency, accountability, and protections to ensure AI delivers on the possibilities it can offer to promote more fairness and equity while protecting privacy, civil rights, and our democracy. Addressing more significant issues of discrimination and bias through an accountability framework will go a long way toward addressing disinformation. In addition to addressing mis- and disinformation, we believe Congress should enact broader AI protections that ensure:

  • Safety and effective systems are developed with diverse stakeholders to ensure that risks, concerns, and impacts are considered and addressed.
  • Algorithms must be free of discrimination with proactive and continuous measures — that are transparent and trustworthy — to protect the public from discrimination.
  • Consumers should be protected from privacy violations in the design and based on defaults in systems.
  • The public must know AI is being used and understand its impacts.
  • People should be able to opt out of AI and get access to a human being.

The REAL Political Advertisements Act and the Protect Elections from Deceptive AI Act are essential steps in the right direction. Transparency into whether AI has been used in ads will help to keep voters informed, and a prohibition on the distribution of materially deceptive audio or visual media by political campaigns will help ensure the integrity of our democracy. The REAL Political Advertisements Act is a commonsense step forward in addressing some of the problems caused by using AI in political ads and the risks that come with it. We also view the recently introduced bipartisan legislation to ban deceptive AI-generated content in elections as a promising start to ensuring our democracy is protected against tools promoting false or fraudulent information.

As we enter the 2024 election cycle, it is paramount to keep pressure on the platforms to better address voting disinformation and to push for additional, continued oversight of the platforms:

  • Besides consistent enforcement of their rules, platforms must do a better job of addressing non-English speaking disinformation, mainly Spanish and Asian language disinformation, particularly on video-focused platforms such as YouTube and TikTok.
  • We have continually pushed the platforms to provide more data on questionable content regarding voting/elections (as well as civil rights and hate and bias issues) so that we can work together with the platforms to utilize that data and find more solutions to address disinformation.
  • As generative AI advances, the potential for rampant false content and the rapid spread of disinformation is alarming. Platforms and tech companies must have policies, systems, and guardrails to stem the disinformation resulting from generative AI.
    • The Brennan Center has noted that platforms can devote more resources to identifying and removing coordinated bots and labeling deepfakes that could influence elections.[42]
  • The Brennan Center has also suggested that the Cybersecurity and Infrastructure Security Agency should create and share resources to help election offices address disinformation campaigns that exploit deepfake tools and language models to undermine election processes;[43] and, to reduce the risk of AI misuse by political campaigns, the Federal Election Commission should ensure that its political ad disclosure requirements cover the full range of online communications currently permitted under federal law.[44]
  • As Public Knowledge has noted, a whole of society approach is needed to address the issues presented by generative AI and restore trust in our information environment. This can include policymakers creating incentives for the technology platforms to change their policies and product design, and they should foster more competition and choice among media outlets. Civil society should convene stakeholders, including from the communities most impacted by misinformation, to research and design while protecting privacy and freedom of expression.[45]

Congress must use its oversight powers and require the platforms to provide more data and solutions on voting/election disinformation and AI. Disinformation, sometimes driven intentionally by foreign governments in our election cycles, often targets Black and Latino communities and poses significant risks to our society.

To that end, Congress should:

  • Ensure coordinated follow-through by federal agencies using congressional oversight.
  • Launch sustained public engagement with diverse stakeholders through hearings and other fora.
  • Continue oversight hearings and deliberations on AI legislation with a focus on addressing the issues surrounding election disinformation and hate speech.
  • Along with state legislatures, regulate AI to help identify AI-generated content and limit interference with elections.
  • Direct DHS to develop training, tools, and resources to ensure election officials and administrators can detect, address, and prevent abuses of AI in the election context.
  • Press DOJ to continue to aggressively enforce federal voting rights laws and ensure that voters of color are not targeted with disinformation through the use of AI to suppress their vote.

VI. Closing

If we poison our democratic soil with false statements about our own elections and fear that they are rigged, we create ground that truth dies in — but where propaganda, planted by those who seek to deceive, can become invasive using the tools technology provides. That means we must consider whether we are farmers or flame throwers when it comes to our democracy.

We stand ready to work to find solutions that will keep our democracy safe and stop the suppressive effect that AI-generated disinformation can have on civil rights and racial justice. Our civil rights and the integrity of our democracy are at stake.

Thank you for inviting me to testify today. I am pleased to answer any questions you may have.

 

[1] BuzzFeed News, “You Won’t Believe What Obama Says In This Video!,” YouTube (2018), https://www.youtube.com/watch?v=cQ54GDm1eL0.

[2] Christine Fernando, “Election Disinformation Targeted Voters of Color in 2020. Experts Expect 2024 to be Worse.,” Associated Press (July 29, 2023), https://apnews.com/article/elections-voting-misinformation-race-immigration-712a5c5a9b72c1668b8c9b1eb6e0038a.

[3] Jordan Liles, “Photo Shows Dog Peeing on Sign for Biden, Not Trump,” Snopes (Oct. 21, 2022), https://www.snopes.com/fact-check/dog-peeing-trump-sign/.

[4] Report of the Select Committee on Intelligence of the United States Senate on Russian Active Measures Campaigns and Interference in the 2016 U.S. Election, Volume 2: Russia’s Use of Social Media with Additional Views, 116th Cong. (2020) https://www.intelligence.senate.gov/sites/default/files/documents/Report_Volume2.pdf.

[5] Laura Romero, “Colorado Voting Officials Adopt Safety Measures as State Becomes Target for Election Conspiracists,” ABC News (May 9, 2022), https://abcnews.go.com/US/colorado-voting-officials-adopt-safety-measures-state-target/story?id=84546801.

[6] Britt Paris and Joan Donovan, Deepfakes and Cheap Fakes: The Manipulation of Audio and Visual Evidence, Data & Society (2019) https://datasociety.net/wp-content/uploads/2019/09/DS_Deepfakes_Cheap_FakesFinal-1.pdf  (Deepfakes and Cheap Fakes).

[7] Id at 2.

[8] Suzie Dunn, “Women, Not Politicians, Are Most Often Target By Deepfake Videos,” Centre for International Governance Innovation (Mar. 3, 2021), https://www.cigionline.org/articles/women-not-politicians-are-targeted-most-often-deepfake-videos/.

[9] Public Service Announcement, “Malicious Actors Manipulating Photos and Videos to Create Explicit Content and Sextoration Schemes,” Federal Bureau of Investigation (June 5, 2023), https://www.ic3.gov/Media/Y2023/PSA230605.

[10] Deepfakes and Cheap Fakes at 7.

[11] Ayesha Rascoe, “How Real is the Threat of AI Deepfakes in the 2024 Election,” NPR (July 30, 2023), https://www.npr.org/2023/07/30/1190970436/how-real-is-the-threat-of-ai-deepfakes-in-the-2024-election.

[12] Louis Jacobson and Loreben Tuquero, “A Pro-Ron DeSantis Ad Used AI to Recreate Donald Trump’s Voice. Experts Say It Won’t Be The Last.,” Poynter (July 20, 2023).   https://www.poynter.org/fact-checking/2023/desantis-never-back-down-pac-ai-ad-trump-voice/.

[13] Megan Hickey, “Vallas Campaign Condemns Deepfake Video Posted to Twitter,” CBS News (Feb. 27, 2023),  https://www.cbsnews.com/chicago/news/vallas-campaign-deepfake-video/.

[14] Rob Lever, “Deepfake Video of Elizabeth Warren Spreads on TikTok,” AFP Factcheck (Mar. 6, 2023), https://factcheck.afp.com/doc.afp.com.33AE6KT.

[15] Nur Ibrahim, “Did Biden Call for a Military National Draft?,” Snopes (Mar. 2, 2023),  https://www.snopes.com/fact-check/biden-military-national-draft/.

[16] Mekela Panditharatne and Noah Giansiracusa, “How AI Puts Elections at Risk – And the Safeguards Needed,” Brennan Center for Justice (July 21, 2023),  https://www.brennancenter.org/our-work/analysis-opinion/how-ai-puts-elections-risk-and-needed-safeguards (Panditharatne and Giansciracusa).

[17] Id.

[18] Id.

[19] Ben Paviour and Miles Parks, “Virginia Becomes the Latest GOP-Governed State to Quit a Voter Data Partnership,” NPR (May 11, 2023), https://www.npr.org/2023/05/11/1175662382/virginia-eric-withdrawal.

[20] Ethan Magoc, “Flurry of Voter ID Laws Tied to Conservative Group ALEC,” NBC News (Aug. 21, 2012), https://www.nbcnews.com/news/investigations/flurry-voter-id-laws-tied-conservative-group-alec-flna955652.

[21] Molly Jackman, “ALEC’s Influence Over Lawmaking in State Legislatures,” Brookings (Dec. 6, 2013),  https://www.brookings.edu/articles/alecs-influence-over-lawmaking-in-state-legislatures/.

[22] Sarah Childress, “‘Unprecedented’ Number of Restrictive Voting Laws Being Introduced,” PBS (May 31, 2012), https://www.pbs.org/wgbh/frontline/article/unprecedented-number-of-restrictive-voting-laws-being-introduced/.

[23] Andrew Cohen, “No More Pretending: Republicans Admit Vote Restrictions Are All About Winning,” Washington Monthly (Mar. 26, 2021), https://washingtonmonthly.com/2021/03/26/no-more-pretending-republicans-admit-vote-restrictions-are-all-about-winning/.

[24] Arturo Garcia, “Did North Carolina Admit to Target Black Voters with a ‘Voter ID Law?,’” Snopes (Oct. 22, 2018), https://www.snopes.com/fact-check/north-carolina-voter-id/.

[25] “What Research Tells Us About Voter Fraud,” Cornell University (accessed Sept. 25, 2023), https://evidencebasedliving.human.cornell.edu/blog/what-research-tells-us-about-voter-fraud.

[26] Ten Years After Shelby County v. Holder: Charting the Path Forward for Our Democracy, The Leadership Conference Education Fund (June 30, 2023) https://civilrights.org/resource/ten-years-after-shelby-county-v-holder-charting-the-path-forward-for-our-democracy/.

[27] Alex Kasprak, “Did 5.7 Million ‘Illigal Immigrants’ Vote in the 2008 U.S. Election?,” Snopes (Jun. 23, 2017),  https://www.snopes.com/fact-check/illegal-immigrants-2008-election/.

[28] Reuters Fact Check, “Fact Check-Re-examining How and Why Voter Fraud is Exceedingly Rare in the U.S. Ahead of the 2022 Midterms,” Reuters (June 2, 2022), https://www.reuters.com/article/factcheck-fraud-elections-idUSL1N2XP2AI.

[29] Voting Laws Roundup: October 2022, Brennan Center for Justice (Oct. 2022), https://www.brennancenter.org/our-work/research-reports/voting-laws-roundup-october-2022.

[30] Voting Laws Roundup: June 2023, Brennan Center for Justice (June 2023), https://www.brennancenter.org/our-work/research-reports/voting-laws-roundup-october-2022.

[31] Kira Lerner, “Criminalizing the Vote: GOP-Led States Enacted 102 New Election Penalties After 2020,” Governing (July 17, 2022), https://www.governing.com/now/criminalizing-the-vote-gop-led-states-enacted-102-new-election-penalties-after-2020.

[32] Paul M. Barrett, Spreading the Big Lie: How Social Media Sites Have Amplified False Claims of U.S. Election Fraud, NYU Stern Center for Business and Human Rights (Sept. 2022),  https://bhr.stern.nyu.edu/tech-big-lie.

[33] Lindsay Whitehurst and Christina Cassidy, “Election Workers Are Being Bombarded With Death Threats, the U.S. Government Says,” PBS (Aug. 31, 2023), https://www.pbs.org/newshour/politics/election-workers-are-being-bombarded-with-death-threats-the-u-s-government-says.

[34] Jacob Knutson, “Over 140 Threats Against Election Workers Recorded in Maricopa County,” Axios (Nov. 6, 2022),  https://www.axios.com/2022/11/06/arizona-maricopa-county-threats-against-election-workers.

[35] Leadership Conference and Democracy Groups Urge Social Media Platforms to Address Voter Disinformation Ahead of Midterms, The Leadership Conference on Civil and Human Rights (Nov. 9, 2022), https://civilrights.org/resource/leadership-conference-and-democracy-groups-urge-social-media-platforms-to-address-voter-disinformation-ahead-of-midterms-2/; The Leadership Conference and 120 Civil Rights & Democracy Groups Urge Social Media Platforms to Take Meaningful Steps to Address Election Disinformation, The Leadership Conference on Civil and Human Rights (May 12, 2022), https://civilrights.org/resource/the-leadership-conference-and-120-civil-rights-democracy-groups-urge-social-media-platforms-to-take-meaningful-steps-to-address-election-disinformation/.

[36] Civil Rights Concerns Over YouTube Decision to Allow Election Disinformation, The Leadership Conference on Civil and Human Rights (June 26, 2023), https://civilrights.org/resource/civil-rights-concerns-over-youtube-decision-to-allow-election-disinformation/

[37] https://www.nytimes.com/2023/02/14/technology/disinformation-moderation-social-media.html

[38] Mike Wendling, “Musk Claims Zapping Bots Will Stop False Information,” BBC (Apr. 12, 2023), https://www.bbc.com/news/live/world-us-canada-65247272/page/2.

[39] Ben Collins, Brandy Zadronzy, and David Ingram, “Days Before the Midterm, Twitter Lays Off Employees Who Fight Misinformation,” NBC News (Nov. 4, 2022), https://www.nbcnews.com/tech/misinformation/twitter-fires-employees-fight-misinformation-midterm-elections-rcna55750.

[40] Hayden Field and Jonathan Vanian, “Tech Layoffs Ravage the Teams That Fight Online Misinformation and Hate Speech,” CNBC (May 27, 2023), https://www.cnbc.com/2023/05/26/tech-companies-are-laying-off-their-ethics-and-safety-teams-.html.

[41] Blog, “An Update to Our Approach to US Election Misinformation,” YouTube (June 2, 2023), https://blog.youtube/inside-youtube/us-election-misinformation-update-2023/.

[42] Panditharatne and Giansciracusa.

[43] Id.

[44] Id.

[45] Lisa Macpherson, “Lies, Damn Lies, and Generative Artifical Intelligence: How GAI Automates Disinformation and What We Should Do About It,” Public Knowledge (Aug. 7, 2023), https://publicknowledge.org/lies-damn-lies-and-generative-artificial-intelligence-how-gai-automates-disinformation-and-what-we-should-do-about-it/.