Regulating Artificial Elections

Belmont Law Review Belmont Law Review

Regulating Artificial Elections Regulating Artificial Elections

Courtney A. Barclay Jacksonville University College of Law

Article 2

COURTNEY A. BARCLAY*

intelligence

This Article examines how artificial is reshaping election discourse and testing First Amendment doctrines rooted in the “marketplace of ideas.” It traces the emergence of generative tools, deepfakes, and chatbots in political communication and evaluates whether disclosurebased campaign-finance laws adequately address AI-driven distortions. Surveying the Federal Election Commission’s 2024 interpretive rule on fraudulent misrepresentation, new state laws mandating disclaimers or prohibiting synthetic media, and pending the Article highlights the constitutional and practical shortcomings of current frameworks. It argues that narrowly drawn legal rules must be complemented by self-regulation and technological transparency measures. Drawing from the AdChoices model in data privacy, it proposes a hybrid approach that combines enforceable AI-content labeling with platform-level interventions to inform voters in real time that messages were created or edited with AIrestore electoral supported transparency and public trust without undermining robust political speech.

federal proposals,

tools. The goal

to

is

II. III.

INTRODUCTION ........................................................................................... 66 CAMPAIGNS IN THE AGE OF AI ...................................................... 70 I. A. Definitions ............................................................................... 71 B. AI in Elections ........................................................................ 72 THE MARKETPLACE OF IDEAS ....................................................... 74 CAMPAIGN FINANCE REFORM AS A MARKET CORRECTOR ........... 77 A. Foundations of Campaign Finance Law ................................. 79 A. Bipartisan Campaign Finance Reform Act ............................. 81 B. A Reliance on Disclosures and Disclaimers ........................... 83 D. Campaign Finance Market Correctors and First Amendment ............................................................................. 86

IV.

1. Prevention of Corruption in the Electoral System ............ 87 2. Political Equality in the Electoral System ........................ 90 AI DISCLOSURES AND PROHIBITIONS ............................................ 94 A. States Look to Curb Impact of AI in Elections ....................... 95 B. Federal Proposals Track State Efforts ..................................... 97 V. A HYBRID APPROACH TO AI POLICY ............................................ 98 CONCLUSION ............................................................................................. 100

Taylor Swift and Lady Gaga endorsed former President Donald Trump. Ryan Reynolds was photographed wearing a Kamala Harris T-shirt. Shaquille O’Neal threw Tim Walz out of his restaurant. J.D. Vance’s dog is actually “rented” for the campaign. None of this is true, but these examples represent some of the viral and false 2024 election social media posts created using artificial intelligence tools.1 These tools have become increasingly accessible and popular and are expected to play a more significant role in online communication, including political campaigning.

Although the 2024 election cycle disappointed some who predicted an explosive use of AI to deceive voters,2 there were key moments punctuated by deepfake videos, photos, and audio files that foreshadow how AI could sway public opinion and disrupt elections.3 Social media is one of the primary sources of news for Americans, with more than half reporting that they get their news from these platforms.4 In particular, younger audiences turn to social media for information about local politics.5 Most Americans were very concerned that artificial intelligence would be used to unduly influence the election through the distribution of manipulated and falsified content,6 and have called for the government and technology companies to censor fake news.7

in First Amendment

This growing desire for more regulation of misinformation runs counter to longstanding free speech principles, particularly the counter speech doctrine—enshrined jurisprudence and scholarship—that states that the remedy for “falsehoods and fallacies” is not censorship, but more speech.8 This enduring approach trusts audiences to evaluate competing ideas to reach rational, informed decisions and is embodied in the marketplace of ideas theory of free expression invoked by Oliver Wendell Holmes in a 1919 dissenting opinion to explain the value of free speech and unfettered debate.9

It was fifty years later when this marketplace theory emerged in the United States as a dominant approach to free speech.10 Under this model, courts have consistently limited infringement on the freedom of expression, reasoning that false or harmful speech is the lesser evil to a tyrannical suppression of ideas—that ardent protection of speech, free from government intrusion, is essential for effective self-governance and accountable limited government.11 Largely developed when debates were waged in print or on soapboxes, the marketplace model relies on assumptions of access, consideration of competing ideas, and rationality of the public.12

This ability of the people to effectively engage in the free trade of ideas is often undermined by real-world practicalities that result in potential imbalances, allowing powerful speakers to overshadow speakers with fewer resources.13 These imbalances may promote the spread of false and even harmful speech, and the courts have upheld government regulations on speech when they act as market corrections to address “real world

conditions [that] also interfere with the effective operation of the marketplace of ideas.”14

Although modern technological developments have, in many ways, enabled greater public access and engaged communication, these power dynamic concerns have also been exacerbated by new media.15 Seemingly omniscient social algorithms encourage and enable users to consume information that, at best, confirms their biases and limits or eliminates exposure to different perspectives—creating an echo chamber despite the myriad speakers publishing online—and, at worst, proliferates false and damaging information that maliciously manipulates the discourse.16

This kind of disruption to the idealized marketplace of ideas presents a significant risk in the context of democratic elections. This is somewhat at odds with the Court’s approach of protecting speech most stringently when the speech is political. However, the Court has found the integrity of U.S. elections to be a compelling reason to uphold market corrections, specifically campaign finance reform laws.17 These laws restrict contributions to candidates, political parties, and political committees; limit sources of funding for political ads aired on TV and radio in proximity to election day; and mandate disclosures and disclaimers to identify the sources of funding for political ads and campaign activities.18 These three approaches attempt to address the potential distortion of the marketplace caused by influxes of resources poured into federal elections through large campaign donations and independent expenditures by individuals and organizations intending to influence the outcome of an election.

Most recently, Congress, the Federal Elections Commission (the “FEC”), and state governments have turned their attention to artificial intelligence (“AI”), which, for all its potential benefits, is increasingly autonomous and unpredictable in its interactions.19 AI can promote misinformation and hinder the public’s ability to make informed choices and engage in rational discourse.20 While it is the conceit of every era to

believe that the circumstances of the time are unprecedented, the disruption, uncertainty, and unease caused by AI seem extraordinary, if for no other reason than the speed and extent of the changes it is generating.21 Chatbots, deepfakes, and targeted messaging—all powered by various forms of AI— are being used to manipulate the marketplace of ideas and disadvantage the public, particularly in their role as voters.22 AI-generated robocalls impersonating candidates with information damaging to their campaign, deepfake videos offering false endorsements of candidates, and deceptive chatbots sending voters to nonexistent polling places are examples of the influence AI-generated speech can exert in election discourse.23

Thus far, the federal government has failed to enact a regulatory solution, although several were introduced to the 118th Congress, and the 119th Congress is continuing to debate regulatory approaches.24 In 2024, the FEC declined to engage in new rulemaking to specifically address the concerns raised by AI election influencers, opting instead to approve an interpretive rule clarifying that AI-generated content falls under existing regulations prohibiting “fraudulent misrepresentation.”25 State legislatures have acted more quickly, with laws in more than twenty states that primarily focus on disclosure requirements of AI-generated speech by a candidate, and penalties ranging from fines to incarceration26. Some states have legislation pending that would enact a total ban on any “synthetic media” in electioneering communications.27 In 2025, a federal bill put a question mark on all of these state efforts, proposing a ten-year moratorium

on states and cities passing new reforms or enforcing laws regulating AI systems.28

This Article reviews federal proposals and state laws, analyzing the constitutionality and potential effectiveness of regulations of AI-generated election speech, as well as considering and evaluating non-legislative options that may be less restrictive. Section I explores the use of AIgenerated content in elections. Section II reviews the marketplace of ideas theory and its relationship to campaign finance law. Section III discusses current federal campaign finance laws, which govern communications related to the elections for federal office. Section IV discusses the state laws and proposed federal legislative solutions to address the potential harm of AI-generated content. Section V concludes that current legal protections are inadequate to address the harms and recommends a hybrid approach that would enhance the free flow of information, acting as a market correction to the distortion proliferated by AI and other digital tools.

I.

The Court––in elevating disclosure requirements as a primary measure against corruption––has placed heavy emphasis on a marketplace of ideas model that envisions informed consumers who can ably evaluate content they consume.29 Artificial intelligence systems are becoming an integral part not only of how we find that content, but also of the content creation itself.30 Generative AI has a vast capacity to impact our daily lives, including healthcare, customer service, education, transportation, and legal services.31 AI is also beginning to change the way political campaigns are run, impacting the information voters consume in media and influencing the people the voters choose to represent them in government, and it has the potential for even greater impact.32

The FEC reported a 260% increase in digital ads between 2014 and 2018, and market reports predict a 156% increase in digital campaign ads in

2024 over the spending in 2020.33 Candidates have not been quick to adopt AI in their campaigning in 2024—at least not in voter-facing activities—in large part due to internal polling indicating that voters distrust the technology.34 However, more than half of Americans—as reported in a poll conducted by that individuals and organizations will use AI to create and distribute false information that will influence the outcome of the election.35

the Pew Research Foundation—are concerned

A. Definitions

The term “artificial intelligence” has been used to encompass a wide variety of computerized systems with varying abilities. The common thread in scholarly definitions is that AI systems can “think and act rationally in ways that mirror human behavior and intelligence.”36 These systems often incorporate machine learning, which enables the program to enhance its performance based on experience, or deep learning, which enables the program to learn from and analogize millions of pieces of data.37

The origin of the term “artificial intelligence” can be traced to a 1956 conference, the Dartmouth Summer Research Project on Artificial Intelligence.38 It has come to the forefront of technology policy–as well as countless professions–recently because, for the first half century after the conference, much of the work in AI was focused on theory, not practice.39 The development of more powerful computers, large data sets, and

improved algorithms catalyzed applications of AI.40

the advancement of

the practical

One of the developments in AI applications that has opened the field to more diverse users is generative AI, such as ChatGPT, a chatbot developed by OpenAI that uses deep learning AI to generate responses based on user prompts.41 This technology is being used in a variety of capacities, from everyday Internet searches to advancing fusion technology research.42 These generative AI systems can be used to create synthetic media in the form of videos, images, and audio files, with the potential of mass disinformation, raising significant concerns about the use of this technology in elections around the world.43 In fact, OpenAI CEO Sam the U.S. Senate Judiciary Altman said, when he subcommittee meeting in 2023, mass disinformation in elections was “one of [his] areas of greatest concern.”44

testified before

B. AI in Elections

This seemingly boundless capacity has raised concerns about the potential harms of systems like ChatGPT, including the proliferation of mis/disinformation, deepfakes, and persona bots “hacking” political systems.45 Researchers at Stanford University’s Polarization and Social Change Lab and the Institute for Human-Centered Artificial Intelligence studied the use of AI in political persuasion.46 The experiment demonstrated the ability of AI to persuade individuals on policy issues such as gun control, tax credits, and smoking bans.47 The authors found that compared to the human authors, participants rated the author of AI messages as being more factual and logical, but less angry, unique, and less likely to use storytelling, and that AI-generated messages were as persuasive as messages crafted by lay humans.48

This study confirmed the finding from a 2019 experiment in which Max Weiss––a student at Harvard––identified the vulnerabilities of federal comment processes, especially as related to AI-generated comments.49 Weiss coded a bot that generated more than 1,000 comments to the Medicaid.gov website and then recruited participants to evaluate which of the submitted comments were crafted by AI as opposed to the legitimate comments submitted by individuals.50 Weiss found that participants accurately identified the use of AI less than half the time.51

Although the 2024 election was the first federal election in the United States since the introduction of ChatGPT, chatbots and AI are not new to politics. In 2016, approximately 20 percent of all political messages posted on Twitter were posted by bots,52 and foreign operatives used fake social personas, powered by chatbots, to post content related to racial issues to influence the presidential election53 In 2018, leading up to the midterm elections, approximately half of the online discussion of the immigrant caravan was authored by chatbots.54 Reportedly, the Trump and Biden campaigns spent $200 million on AI-driven Facebook advertisements in 2020.55

In January 2024, the Cybersecurity and Infrastructure Security Agency issued a report detailing the risks associated with the use of AI in the 2024 election cycle.56 The report details concerns with AI being used to manipulate audio, image, text, and video content to influence and disrupt the election process, offices, officials, and vendors.57 Specific concerns

identified by the study were the proliferation of misinformation and the delegitimization of the election.58

Experts in technology and politics were interviewed in a study by researchers at the University of Texas Center for Media Engagement.59 The experts revealed that local elections are ground zero for AI campaigning, and synthetic media is being readily deployed in these races as effective disinformation, likely because it is harder for the public to detect misuse against these lesser-known candidates than against well-known political figures vying for federal office.60 Further, even in more high-profile campaigns, generative AI is being used to target voters, engage in fundraising, and draft speeches.61

II.

The concept that a free exchange of ideas is a necessary prerequisite for self-government was introduced into U.S. jurisprudence in a dissent by Justice Oliver Wendell Holmes in Abrams v. United States.62 The Abrams Court found that the publication of dissident materials during a time of war was not protected expression under the First Amendment.63 Holmes dissented from the majority, arguing that such restriction would hamper the debate on public policy, thereby undercutting the search for truths:

[W]hen men have realized that time has upset many fighting faiths, they may come to believe even more than they believe the very foundations of their own conduct that the ultimate good desired is better reached by free trade in ideas-that the best test of truth is the power of the thought to get itself accepted in the competition of the market, and that truth is the only ground upon which their wishes safely can be carried out. That at any rate is the theory of our Constitution.64

This marketplace analogy the background of U.S. jurisprudence until Justice William J. Brennan revived it in 1965.65 In

languished

Lamont v. Postmaster General, the Court determined that a statute mandating the detention of unsealed mail determined to be “communist political propaganda” was inconsistent with the First Amendment.66 Brennan, in a concurring opinion, emphasized that the implementation of this law would be to restrict people’s ability to receive information, a fundamental right.67 Brennan argued that without protecting the recipient’s rights to receive and consider ideas, the First Amendment protections for dissemination would be meaningless: “[i]t would be a barren marketplace of ideas that had only sellers and no buyers.”68 Only two years after Brennan’s initial revival of the marketplace metaphor, it found general acceptance in the Court69 and has been a fixture in free speech cases for more than fifty years.70

Critics of the marketplace theory have contended that it does not adequately protect free speech or promote the desired search for truth.71 In 1967, the same year that the marketplace metaphor first gained majority approval in the Court, Jerome A. Barron criticized the theory in a Harvard Law Review article.72 Barron criticized the Court for its blanket protection of “free speech” through a prevention of government intervention.73 Barron averred that by focusing on the free flow of information, the Court has been indifferent to the fact that communication channels had changed and that access to those channels was limited to certain interest groups; this imbalance of access allowed powerful speakers to overshadow those with fewer resources.74 “The ‘marketplace of ideas’ view has rested on the assumption that protecting the right of expression is equivalent to providing for it.”75 It is not the government that may “most effectively abridge free expression,” but the media.76 The marketplace of ideas model, according to Barron, no longer protected robust debate in the age of media control because it restricted the government from ensuring equal access to the debate.77

Barron argued that restraining the government would not ensure free speech.78 Rather, he would have required a restraint on “private

groups” that controlled the mass media.79 He advocated a balance of the interests of the owners of media outlets and members of the public seeking a “forum in which to express their point of view.”80 To ensure equal access to the mass media, Barron suggested a constitutional right of access to the media.81 This right of access calls for the “burial” of the classic marketplace of ideas:82

the development of private restraints on free With expression, the idea of a free marketplace where ideas can compete on their merits has become just as unrealistic in the twentieth century as the economic theory of perfect competition. The world in which an essentially rationalist philosophy of the first amendment was born has vanished and what was rationalism is now romance. 83

C. Edwin Baker, echoing concerns of critics like Barron that the monopolization of media channels led to a failure of the marketplace, also argued that the classic marketplace of ideas model rests on certain faulty assumptions: (1) truth is objective, and (2) people are rational and able to perceive the truth.84 Baker said that truth is constantly shaded by experience and, therefore, is not—and cannot be—objective.85 Proving the faulty nature of the first assumption, Baker argued, also discredits the second; if there is no objective truth, people cannot discern it.86 Further, Baker argued that even if there is a discernible truth, people cannot look past the packaging of the message.87 Emotions greatly impact people’s decisions and perceptions; they will not be able to see past form and frequency to the “real” information.88 Thus, Baker concluded that the classic marketplace of ideas fails to produce the value intended by the metaphor.89

Baker reviewed several proposals for a revised marketplace model.90 These models ranged from advocacy subsidies to expenditure limitations to creating free media for speech.91 However, in his evaluation of these proposals, Baker found them all unacceptable for First Amendment doctrine.92 Rather, he proposed the Liberty Model, which would expand

First Amendment doctrine beyond the marketplace model to include a focus on autonomy and self-fulfillment.93

These critiques of the marketplace, particularly Barron’s rebuke of the limited access in modern media channels, have been considered but not adopted widely by the Court. In Red Lion Broadcasting Company v. Federal Communications Commission, the Court upheld the federal requirements for broadcasters to provide public access to broadcast outlets as a forum for speech, noting that, in that case, regulation was necessary to promote a robust marketplace of ideas and ensure speakers could reach through particular media.94 Although Red Lion upheld audiences regulations protecting the speaker’s right of access, much of the Court’s marketplace jurisprudence focuses the analysis on the right of the recipient to receive information.95 In election cases, the Court has been particularly concerned with preserving the integrity of the election system and ensuring voters have access to information about candidates and issues.

III.

The importance of money in U.S. politics has drastically increased since the first elections following the American Revolution.96 Campaigns funded entirely by the candidates were quickly abandoned as costs for campaigning escalated from a few hundred dollars to thousands and hundreds of thousands.97 In 2024, the total cost for the federal election was $9.5 billion for combined congressional races and $5.3 billion for the presidential race; incumbents running for the U.S. House of Representatives raised, on average, more than $3 million.98

As costs rose, corporations and large independent donors became more important in the electoral process, and concern grew over the influence of these donations.99 State legislatures initiated total bans on corporate donations to candidates.100 In 1907, the federal government

banned corporate contributions and spending for federal election campaigns.101 For more than 100 years, Congress has continued to regulate federal elections, placing limits on contributions and spending, mandating the disclosure of campaign spending and donations, and requiring disclaimers on advocacy communications.102

Campaign finance law regulates money in politics in three ways: contribution limits, prohibitions on types of donors, and disclosure and disclaimer requirements. Contribution limits include upper limits on the amount individual donors can give to candidates, parties, and political committees.103 Prohibitions on types of donors include restrictions on donations from corporations, unions, and foreign nationals.104 Disclosure requirements mandate reporting of contributions and expenditures by candidates, political committees, and other organizations or individuals who make expenditures or electioneering communications.105 Disclaimer requirements are mandatory attribution statements, which include funding sources and, when applicable, candidate approvals.106

Since the first comprehensive campaign finance reform effort in the 1970s, the major amendments have been aimed at closing loopholes

exposed by new interests, technologies, and communication platforms.107 Despite these regulations, campaign spending and concern over undue influence in the electoral system have continued to rise at considerable rates.108 Increasingly, lawmakers and judges alike have placed more and more faith in disclosure and disclaimer requirements instead of imposing stricter limits on spending.109 This dependence is rooted in the idea that the solution to harmful speech is more speech.110 However, disclaimer laws have not kept pace with evolving communication platforms or strategies.111 This Section provides a review of the history of campaign finance laws, with emphasis on the disclaimer mandates.

A. Foundations of Campaign Finance Law

Campaign finance laws regulate the flow of money in federal campaigns to protect against corruption by excessive contributions. The formula of campaign finance laws changed over the years when new challenges arose. Federal campaign finance laws originated in the early twentieth century when corporate donations to presidential campaigns were a transparent method of “curry[ing] influence” with the administration.112 In response, the Tillman Act of 1907, one of the earliest attempts at campaign finance reform, prohibited corporations from donating to federal political campaigns.113

the following half-century, Congress created disclosure requirements, spending limits, and contribution limits to ward off further threats of corruption.114 With the advent of television and the Watergate era, Congress embarked on its first attempt at a comprehensive reform, resulting in the Federal Election Campaign Act of 1971 (“FECA”). The purpose of the Act was “to promote fair practices in the conduct of election campaigns for Federal political offices.”115 FECA placed limits on contributions and expenditures—indirect gifts of value—by corporations and candidates in

connection with federal elections.116 However, FECA did not establish an independent oversight agency, and the 1972 elections led to more than 7,000 reports to the Department of Justice of campaign finance abuse, contributing to significant amendments in the FECA in 1974.117

The 1974 amendments established the FEC as the sole agency responsible for overseeing the administrative functions of the campaign finance system.118 Congress also enacted limits on both contributions and expenditures that applied to federal candidates and political committees, as well as disclosure requirements for candidates’ contributions and expenditures.119 The constitutionality of the 1974 amendments was challenged by Senators James L. Buckley and Eugene McCarthy in the aforementioned Buckley v. Valeo.120

In response to those challenges, the Supreme Court issued a per curiam opinion in Buckley v. Valeo121 that has shaped all subsequent campaign finance reform. The Court attempted to fence in campaign finance reform to protect First Amendment rights implicated by some of the law’s provisions.122 The Buckley Court held that limits on candidates’ expenditures and “independent” expenditures by individuals and groups infringed on First Amendment rights.123 were unconstitutional and However, limiting contributions and “coordinated expenditures,” the Court found, satisfied the compelling government interest in reducing corruption of the election system.124

Further, the Buckley Court found disclosure requirements directly advanced the government’s interests of providing the electorate with information, deterring corruption, and detecting violations of FECA limits

on contributions.125 Despite recognizing a potential for donors to face public scrutiny, the Court held that the disclosure requirements “appear[ed] to be the least restrictive means of curbing the evils of campaign ignorance and corruption that Congress found to exist.”126

Following this decision, Congress revised the law to allow for the constitutionally protected independent expenditures.127 However, Congress also enacted additional disclosure requirements to ensure reporting of these activities.128 These post-Buckley amendments created the structure of U.S. campaign finance law that stood for thirty years: limits on contributions, protected speech through independent expenditures, and strong reporting requirements for campaign and party contributions and spending.

A. Bipartisan Campaign Finance Reform Act

Nearly thirty years passed before Congress again significantly amended FECA. In 2002, Congress passed the Bipartisan Campaign Finance Reform Act (“BCRA”) to “close” some of the loopholes that the FECA left open.129 Specifically, the sponsors of the bill, Senators John McCain and Russell Feingold, expressed concern with unregulated softmoney donations—money raised outside the scope of FECA.130 These types of donations had allowed contributors to bypass the FECA’s contribution

and expenditure limits for years.131 In addition to addressing concerns over soft-money, provisions in BCRA included increased contribution limits and limits on communications leading up to an election.132

To mitigate soft-money donations and expenditures, Congress set limits on the activities of office holders, candidates, and national parties in relation to fundraising.133 Congress also amended the definition of “public communications,” which determines, in large part, the scope of the campaign finance legal requirements mandated by FECA and BCRA for campaign communications.134

to

“Federal election activity” was defined

include public communications–“a communication by means of any broadcast, cable, or satellite communication, newspaper, magazine, outdoor advertising facility, mass mailing, or telephone bank to the general public, or any other form of general public political advertising”135—that clearly identifies a federal candidate and either promotes or opposes a candidate for that office.136 Congress restricted the funding for “federal election activity,” requiring that these activities be paid for with federal money only—money that is subject to the restrictions of FECA.137 Public communications funded by anyone other than a candidate must carry disclaimers on the payment of the communications and whether the candidate supported the communication, and these disclaimers must be “clear and conspicuous.”138

Additionally, Congress significantly changed the campaign finance landscape when it created a category of public communications called “electioneering communications,” which Congress defined as “any broadcast, cable or satellite communication” that refers to a clearly identified candidate for federal office, is publicly distributed shortly before an election for the office that candidate is seeking, and is targeted to the relevant electorate.139 BCRA prohibited corporations and labor unions from making or funding electioneering communications not targeted to their internal audiences (e.g., a labor union’s membership or a corporation’s

employees or stockholders).140 Incorporated organizations formed for the purpose of affecting the outcome of elections (known as “527s” for the tax code under which they are organized) are also prohibited from making electioneering communications.141 BCRA allowed individuals to make electioneering communications, provided that no corporate or labor organization funds were used.142

they could prove

Congress intended the electioneering communications restrictions to curb circumvention of spending prohibitions on corporations and unions.143 However, in 2010, the Supreme Court invalidated these prohibitions, placing further emphasis on disclaimer and disclosure requirements to mitigate corporate influence in elections.144

B. A Reliance on Disclosures and Disclaimers

As mentioned above, the Court first reviewed the constitutionality of BCRA in McConnell v. Federal Election Commission, rejecting First Amendment challenges to the disclosure and disclaimer requirements in BCRA and showing great deference to the congressional findings of threats of corruption.145 The Court reiterated Buckley, noting the values served by disclosure: “providing the electorate with information, deterring actual corruption and avoiding any appearance thereof, and gathering the data necessary to enforce more substantive electioneering restrictions.”146

Disclosure and disclaimer requirements—arguably a minor aspect of BCRA—were elevated significantly when the Court invalidated other restrictions on electioneering communications in Citizens United v. Federal Election Commission147 and the limits on aggregate contributions in McCutcheon v. Federal Election Commission.148 The Court found those solutions to political corruption inadequate and advocated that disclosure requirements serve as a primary mechanism to combat corruption in politics, especially because of the impact they can have in online communications, which was not fully considered when BCRA was enacted.149

thus,

filings;

When reporting requirements were initially advanced, information about the election process was effectively shielded from the public.150 The Court in McCutcheon noted that the public would have to pull physical these documents “were virtually copies of FEC inaccessible.”151 With the adoption of digital reading rooms and data-rich websites, many FEC filings are readily available to the public online.152 The filing documents are further “supplemented” by organizations, such as OpenSecrets.org and FollowtheMoney.org, that provide this data to the public in digestible analyses.153 The Court suggested that disclosure may now be “effective to a degree not possible at the time Buckley, or even McConnell, was decided.”154 The accessible and timely disclosure of expenditures can better provide shareholders and citizens with information needed to hold corporations and elected officials accountable for their positions and supporters. Shareholders can determine whether their corporation’s political speech advances the corporation’s interests, and citizens can see whether elected officials are “in the pocket” of so-called moneyed interests.155

According to the Citizens United Court, disclosure and disclaimer requirements are critical because “transparency enables the electorate to make informed decisions and give proper weight to different speakers and messages.”156 These requirements are intended to ensure voters know the source of the message they are receiving so they can better evaluate the weight of the message without the government limiting the quality or amount of political speech available. Unfortunately, the Internet, and particularly social media, has not lived up to the promise the Court articulated. Rather, it has allowed for the spread of misinformation, magnified by the use of chatbots and artificial intelligence.157

Campaign finance disclosure requirements in the United States are designed to promote transparency by mandating that political candidates, parties, and certain organizations publicly report their funding sources and expenditures. This transparency aims to inform voters about the financial influences behind political campaigns, thereby enhancing the democratic process. The U.S. Supreme Court has consistently upheld disclosure

requirements, recognizing their role in educating voters about the sponsors behind political messages.158

these

Despite

intentions,

the effectiveness of disclosure requirements faces challenges. Instances of fraudulent disclosure statements have been reported, where entities obscure the true sources of campaign contributions. For example, during the 2012 election cycle, a donor to a super political action committee (PAC) concealed his identity by using a limited liability company (LLC) formed solely for that purpose, and, in 2024, the Save Western Culture super PAC was accused of using a shell organization, Stop China Now, Inc., to conceal the true sources of its funding, which was used to target Republican candidates. Such practices undermine the transparency that disclosure laws seek to ensure.159

their effectiveness. While

The limitations of information available in disclosure statements further complicate the Federal Election Campaign Act sets forth statutory provisions governing campaign finance,160 there are concerns that current disclosure requirements may not fully capture the complexities of modern campaign financing. Campaigns and PACs frequently use intermediaries, such as LLCs, nonprofit organizations, or other financial entities, to make it more difficult to trace the true source of funding. For example, "dark money" groups—politically active nonprofits that are not required to disclose their donors—can funnel significant amounts into campaigns while shielding the identities of their contributors. According to a 2024 report by OpenSecrets, over $1 billion in dark money was spent in recent election cycles, raising serious concerns about the influence of undisclosed donors on American politics. 161 The prevalence of undisclosed political spending, often referred to as "dark money," has raised concerns about the integrity of disclosed information and elections more generally.

The accessibility of campaign finance disclosure data presents significant challenges. Although this information is publicly available, it is often presented in formats that are difficult for the average citizen to access or interpret. Complex filing systems and inconsistent reporting standards

between states further complicate efforts to create a clear and comprehensive picture of campaign finances.162 Additionally, inaccuracies in reported data diminish its utility. Donor information, such as occupation and employer, is sometimes listed inaccurately or incompletely. An analysis by OpenSecrets highlighted instances where contributors provided unconventional or incorrect descriptions for their occupations, leading to data that is less informative and potentially misleading.163

D. Campaign Finance Market Correctors and the First

Amendment

The Supreme Court has warned that campaign finance regulations “operate in an area of the most fundamental First Amendment activities” affecting “discussion of public issues and debate on the qualifications of candidates,” which is “integral to the operation of the system of government established by [the] Constitution.”164 From this perspective––declared in the first modern campaign finance case––the Court has attempted to reconcile the First Amendment and campaign finance reform laws.

In reviewing First Amendment challenges, the Court, applying “exacting scrutiny,”165 has developed a jurisprudence riddled with divisive opinions.166 Key cases have often been decided by 5-4 votes and are habitually confused by plurality or partial opinions and multiple concurring and dissenting opinions, but the vital importance of an informed electorate and a robust marketplace of ideas.167

they consistently recognize

Regulations of contributions and expenditures have been touted as “inconsistent” with that marketplace by effectively “taking from rich speakers for the benefit of poor ones.”168 The Supreme Court agreed in

Buckley v. Valeo, writing that “the concept that government may restrict the speech of some elements of our society in order to enhance the relative voice of others is wholly foreign to the First Amendment.”169

Supporters of reform often attempt to counterbalance the First Amendment interests with one or more general themes. Scholar Kathleen M. Sullivan identified several themes prevalent in the campaign finance debate: political inequality in voting; distortion, corruption, or political inequality in representation; carpet bagging; diversion of legislative and executive energies; quality of debate; and lack of competitiveness.170 The most common themes cited by reform advocates and addressed by the Supreme Court are prevention of corruption and political equality.

1. Prevention of Corruption in the Electoral System

far

thus

interests

the only

The Supreme Court has said that “preventing corruption or the legitimate and compelling appearance of corruption are government restricting campaign finances.”171 This corruption is generally portrayed as a quid pro quo scenario in which campaign contributions are exchanged for promises of political favors, such as support for or opposition to a piece of legislation.172 There is opposition to the continued use of the anti-corruption rationale, both from practical standpoints and from conceptual outlooks on campaign finance reform.

identified

for

One of the main criticisms of the anti-corruption rationale is the lack of clear criteria for identifying “corruption.”173 Definitions proposed by scholars can be grouped by their basis in legality, public interest, or public opinion.174 The legality-based definition confines corruption to the violation of formal laws or rules, which, according to John Peters and Susan Welch, is “simultaneously too narrow and too broad in scope; all illegal acts are not necessarily corrupt and all corrupt acts are not necessarily illegal.”175 The public interest definition finds corrupt any act that “violates responsibility

toward at least one system of public or civic order.”176 This definition allows for the possibility that an act, while illegal, may actually benefit the political system.177 The difficulty with this type of definition is that it requires, first, a definition or delineation of “public interest.”178 The third category of corruption definitions is based on public opinion–that is, acts are judged to be corrupt or not by the public.179 This approach necessitates the estimation of public sentiment but allows for a continuum of corruptness.180

As Peters and Welch found each of these categories to be insufficient for identifying corruption, they proposed a method of analyzing the components of a potentially corrupt act: the public official involved, the favor provided by the official, the payoff gained by the official, and the donor of the payoff or the recipient of the favor.181 Each of these elements can be assessed using a model for “more corrupt” and “less corrupt” to identify where the behavior falls in the spectrum.182

Additionally, some scholars doubt the actual presence of or potential for corruption in the electoral system. Many large donors hedge their bets—giving to both major political parties.183 Kathleen Sullivan argued that this bet hedging demonstrates a “weak level of confidence in [the donors’] ability to obtain results from any particular beneficiary of their contributions.”184 In fact, congressional behavior suggests that there is a low correlation between contributions and voting behavior.185 Rather, congressmen vote along party lines.186 Although donors may be repaid in

less formal methods, the claimed corruption of the electoral system seems less certain than originally proposed.187

Further, the anti-corruption rationale assumes that politicians, without large contributions, would consider all constituents equally.188 However, groups of constituents that are more organized or reputable in the community often carry more weight with politicians because they are able to mobilize voters.189

The anti-corruption rationale was not fully explicated in Buckley and has evolved to the point of being a barrier for campaign finance law enforcement.190 Initially, the Court deferred to the FEC’s decision-making regarding corruption prevention measures.191 However, in the 1990s, the Court began to question the FEC’s decision-making process.192 In FEC v. National Conservative Political Action Committee, the Court invalidated expenditure limits associated with presidential campaign public funding.193 The Court rejected the FEC’s record demonstrating corruption and the efficacy of the proposed laws in preventing that corruption.194 The Court’s wholesale rejection of the FEC’s factual record left it unclear as to how the FEC could ever show the presence of corruption in a campaign financing scheme.195

Since Buckley, the Court has applied the anti-corruption rationale in a variety of ways. Thomas Burke identified three standards in the Court’s jurisprudence from Buckley to Austin v. Michigan Chamber of Commerce in 1990.196 The first standard that the Court has applied is a quid pro quo type of corruption in which money is donated in exchange for the arrangement of future votes.197 This was the standard originally announced in the Buckley decision, although further examination of the opinions reveals that the Court may have been recognizing a broader scope of corruption.198 One of the possible broader conceptions that Burke identified is monetary influence.199 This idea was expressed by Chief Justice William Rehnquist in FEC v. National Conservative Political Action Committee:

Corruption is a subversion of the political process. Elected officials are influenced to act contrary to their obligations of office by the prospect of financial gain to themselves or infusions of money into their campaigns. The hallmark of corruption is the financial quid pro quo: dollars for political favors.200

Although Justice Rehnquist broadened the environment to include—rather than be defined by—quid pro quo corruption, his opinion struck down expenditure limits, relying on language from Buckley that “the absence of prearrangement and coordination of an expenditure with the candidate or his agent not only undermines the value of the expenditure to the candidate, but also alleviates the danger that expenditures will be given as a quid pro quo for improper commitments from the candidate.”201

Another conception of corruption in the Court has been distortion. Justice Marshall identified this as a different type of corruption from the traditionally acknowledged quid pro quo.202 Distortion is “the corrosive and distorting effects of immense aggregations of wealth that are accumulated with the help of the corporate form and that have little or no correlation to the public's support for the corporation's political ideas.”203 Justice Scalia criticized this as an attempt to revive a justification rejected by the Court in 1974—political equality—under the guise of a “new corruption.”204

2. Political Equality in the Electoral System

Although prevention of corruption is more commonly accepted, it has been argued that corruption is a derivative problem.205 David Strauss contended that corruption actually stems from political inequality.206 The political equality argument suggests that limits on campaign finance contributions promote political equality for voters during an election, equalizing the voices of all citizens.207 The Government in Buckley v. Valeo raised political equality as one of its justifications for limiting contributions and expenditures during elections.208

Kathleen Sullivan explained that this view equates the campaign process to a “kind of shadow election” that results in unequal value of votes.209 However, as Sullivan noted, a serious flaw in this argument is that financing during campaigns is more akin to political speech than it is to casting a vote at the ballot box.210 Generally, political speech is protected from such equalizing restrictions:

[I]n the informal realm of political speech–the kind that goes on continuously between elections as well as during them–conventional First Amendment principles generally preclude a norm of equality of influence. Political speakers generally have equal rights to be free of government censorship, but not to command the attention of other listeners. Under virtually any theory of the justification for free speech, legislative restrictions on political speech may not be predicated on the ground that the political speaker will have too great a communicative impact, or his competitor too little. Conventional First Amendment norms of individualism, relativism, and antipaternalism preclude any such affirmative equality of influence–not only as an end-state but even as an aspiration.211

This argument was proposed to the Supreme Court in 1976 as a compelling government interest for enacting the Federal Election Campaign Act.212 The Federal Election Commission argued that Congress had attempted to “increase opportunities for meaningful participation by ordinary citizens, as voters, supporters, and candidates.”213 However, the Court held that limiting campaign expenditures in an effort to equalize the political discourse during elections was unacceptable.214 “[T]he concept that government may restrict the speech of some elements of our society in order to enhance the relative voice of others is wholly foreign to the First Amendment.”215 The Court further dismissed the political equality argument when it relegated it to the sphere of ancillary arguments, opting to focus instead on the prevention of corruption argument.216 Not all of the justices found the political equality rationale so easy to repudiate. Justice Byron White dissented from the Court’s ban on expenditure caps,

emphasizing the need to “dispel the impression that federal elections are purely and simply a function of money.”217

In addition to the internal strife, scholars criticized the Court’s opinion for failing to recognize the value of political equality. For example, John Rawls wrote:

The Court fails to recognize the essential point that the fairvalue of the political liberties is required for a just political procedure, and that to insure their fair-value it is necessary to prevent those with greater property and wealth, and the greater skills of organization which accompany them, from controlling the electoral process to their advantage. . . . On the Court's view, democracy is a kind of regulated rivalry between economic classes and interest groups in which the outcome should properly depend on the ability and willingness of each to use its financial resources and skills, admittedly very unequal, to make its desires felt.218

The revival of political equality as a compelling interest for campaign finance reform laws has been heavily supported. Senator John McCain, a chief architect of the Bipartisan Campaign Reform Act of 2002, heralded the 2002 reform efforts for reminding the American people that the American government belongs to all citizens, not just those who “can afford enormous payments to parties and candidates.”219 Senator McCain focused on effective participatory self-government as a goal of the reform efforts he spearheaded.220 Judge Calabresi of the U.S. Court of Appeals for the Second Circuit wrote, “The notion that intensity of desire [to support a candidate] is not well-measured by money in a society where money is not equally distributed has been, since Buckley, the huge elephant—and donkey—in the living room in all discussions of campaign finance reform.”221

Since the Buckley decision, scholars have debated the advantages of political equality as a justification for campaign finance reform laws. Noted scholar Cass Sunstein argued, “Insofar as Buckley rejects political equality

as a legitimate constitutional goal, it should be overruled.”222 Richard L. Hasen has argued that political equality is not possible when the media is elevated to “an even more preeminent place than they already have in the shaping of public attitudes toward federal candidates.”223 Jamin Raskin and John Bonifaz contended that political equality is the “key First Amendment issue at stake” because it ensures that all citizens are able to participate in the electoral process.224 However, Bradley A. Smith, former chairman of the Federal Election Commission, averred that political equality is not a right guaranteed by the Constitution and, further, that promoting such equality effectively abridges the right to free speech.225

In the midst of this scholarly debate, the Supreme Court reopened the legal conversation on the equality rationale. Since Buckley, the Court has staunchly refused to uphold any campaign finance reform efforts unless they were purposed to prevent corruption.226 However, the Court’s decision in McConnell v. FEC has been heralded as a shift in campaign finance jurisprudence.227 Even in the aftermath of this proposed shift, the Court has not expressly adopted the political equality argument that was summarily rejected in Buckley.

Leading up to the McConnell decision, the Court provided some foreshadowing to its jurisprudential shift. In 2000, Justice Breyer issued a concurring opinion in Austin v. Shrink Missouri PAC, in which he announced that Buckley’s rejection of political equality as a compelling interest “cannot be taken literally.”228 Breyer noted that there are many times that the First Amendment “permits restrictions on the speech of some in order to prevent a few from drowning out the many.”229 Specifically, Breyer considered the restrictions placed on congressmen for the efficacy and democratization of floor debates as one example.230

The McConnell decision has been heralded as the true turning point for the Court in relation to the political equality rationale.231 Professor Hasen wrote that the majority opinion:

takes pains to show its fidelity to Buckley, tripping over itself to apply the corruption (as anticircumvention) rationale to as many BCRA [Bipartisan Campaign Reform Act] provisions as possible. However, a more natural reading of the more controversial aspects of the joint majority opinion is as a sub silentio acceptance of the participatory self-government rationale.232

In a well-cited note, the Court reaffirms its commitment to “preserving the integrity of the electoral process, preventing corruption, and ‘sustaining the active, alert responsibility of the individual citizen in a democracy for the wise conduct of the government’ [as] interests of the highest importance.”233

IV.

Campaign finance disclosure requirements are foundational to promoting transparency and accountability in elections and facilitating the “active, alert responsibility.” These requirements are perhaps the only element of campaign finance reform left intact, and yet the effectiveness of mandatory disclosures is challenged by fraudulent practices, limitations in the scope of required disclosures, and the resulting public skepticism.

As AI tools become simultaneously more sophisticated and userfriendly, facilitating more widespread use, AI is impacting public discourse in increasingly important and varied ways. This signals a shift in the paradigm of communication because AI is not only being used to create content, but also to determine, in large part, what content users will encounter.234 These types of tools were not contemplated when the foundations of either First Amendment law or campaign finance regulations were being laid, and the growing adoption of AI by individual and corporate speakers is disrupting the traditional expectations and norms of human communication. The types of disclosure restrictions relied on in campaign finance laws to combat more traditional influences on elections may not be enough to adequately cure the emerging harms related to the use of AI in electioneering communications.

Although members of the U.S. Congress have expressed support for federal regulation of AI in electioneering communications,235 no responsive federal law has been enacted. The Federal Election Commission issued an interpretive rule in September 2024, clarifying that existing Federal Election Communication Act prohibitions on fraudulent misrepresentations “apply irrespective of the technology used to conduct fraudulent misrepresentation.”236

The AI Interpretive Rule was in response to a 2023 petition by Public Citizen, requesting a new rulemaking to address the use of AI in campaigning.237 Public Citizen raised concerns of “dirty tricks” that can be exacerbated with AI, resulting in damaging the reputation of candidates and deceiving voters.238 The AI Interpretive Rule was a compromise measure that failed to address questions about the threshold for applying the misrepresentation bans, such as necessary intent, type of false claims, and effect of disclaimers. Moreover, the misrepresentation bans only apply to staff members.239 misrepresentations Misrepresentations by organizations or individuals unaffiliated with the campaigns are not regulated under the prohibitions.240

candidates

their

by

or

Without a federal law or new rulemaking from the FEC, many of the restrictions on the use of AI for election communication originate in state law.

A. States Look to Curb Impact of AI in Elections

More than twenty states have adopted legislation that directly addresses the use of artificial intelligence, including deepfakes, in elections, and an additional nineteen states have proposals pending.241 The state

solutions largely focus on disclaimer requirements,242 although Texas and Minnesota go so far as to prohibit the use of artificial intelligence with the intent to harm a candidate.243 These laws are defined in three ways: the category of speech regulated, the scope of the limitation, and the type of penalty.

One of the first challenges states face in regulating artificial intelligence is defining “artificial intelligence.” While some states use traditional definitions based on the machine-based nature of how content is created,244 others opt for a broader approach, limiting “synthetic media,” “deepfakes,” or “materially deceptive media.”245 This broader approach can include content not generated using AI systems.

Most of the states with laws regulating the use of AI-generated content use disclaimer statements to mitigate the potential harm of the content.246 The requirements are limited by the type of statement required, the proximity of the publication to an election, and the intent required.247 States require disclaimers in the content of the communication stating either that the media has been manipulated or that AI was used in its creation.248 Nine states require disclaimers that the content is altered or fabricated if it is published within ninety days of an election,249 California requires disclaimers if such content is published within 120 days before an election and up to sixty days after an election has concluded.250 Nine states do not place a time restriction on the disclosure but rather require disclosures for any altered or fabricated content if it is done with the intent to harm a candidate or influence the outcome of an election.251

The penalties assessed by the states include civil relief and criminal violations. Fourteen states provide civil remedies, such as injunctive relief,

general and special damages, and civil penalties and forfeitures.252 Two states impose criminal penalties for violation of the requirements, punishable by fines and incarceration.253 Four states include civil and criminal penalties.254

B. Federal Proposals Track State Efforts

Proposals for federal legislation track closely to the state efforts.255 Much of the attention is on transparency, relying on disclaimer statements to identify whether AI or other digital alteration has occurred.256 At least one proposal goes further to create a specific right of action to allow federal candidates to enforce against misleading advertisements.257

The AI Transparency in Elections Act of 2024258 amends FECA and requires disclaimer statements to identify if AI was used to generate any image, audio, or video used in an advertisement. This proposal is, in some ways, broader than the state laws and the FEC rules. The disclaimer requirement attaches to any use of AI in election advertising, rather than being limited to communications made with the intent to deceive or within a certain proximity to the election, and it includes any electioneering communication regardless of the speaker.259 However, some of the state laws apply to any communication,260 and the federal proposal focuses on communications for which placement is paid.261

An earlier proposal, the Protect Elections from Deceptive AI Act,262 focuses on materially deceptive content generated by AI and is more comprehensive than the AI Transparency in Elections Act. This proposed legislation creates a reasonable person standard for defining what is deceptive media and provides a right for subjects of that media to seek injunctive relief.263 The bill also sets up a defined deception as defamation

per se in a libel action.264 These protections would be limited to candidates for federal office.265

None of the state or federal proposals sufficiently addresses the legal and practical problems associated with AI content and the democratic process. Each proposal raises questions about timely enforcement, insufficient deterrent values, and free speech challenges. For example, a federal judge issued an injunction against the enforcement of California’s new law creating civil penalties for deepfakes and similar content that may harm the electorate, citing First Amendment concerns; the law was later invalidated as inconsistent with the Communications Decency Act.266

Narrowly drawn legal solutions must be paired with self-regulation to begin to leave breathing room for innovation and positive use of AI tools while protecting political discourse from undue artificial influence.

V.

The state and federal laws, including the pending federal proposals discussed above, fail to adequately address the concerns raised by the public, lawmakers, and leaders in technology and policy. The reality is that misinformation and disinformation in campaigns have proliferated since the beginning of campaigns. The adoption and advancement of social media and artificial intelligence-powered content creation are exacerbating the concerns related to voter information and public trust in democracy.267 Disclaimers limited to a statement that AI was used or that the media was altered will likely be tacked on as an afterthought, or hidden in plain sight, similar to candidate approvals or side effect warnings on pharmaceutical advertisements. The laws that create civil or criminal penalties get closer to deterring the malicious use of AI, especially by candidates, but fail to correct in real time the information being targeted and delivered to voters in their personal social media and news feeds.

A traditional marketplace of ideas model would advocate for a information without market solution: interference. However, similar to the criticism raised by Barron as he analyzed the power of mass media during the rise of television––controlled by gatekeepers and increasingly consolidated media companies––AI is

the voters evaluate

let

distorting the marketplace with fake social media accounts, content generators that can easily create realistic manipulated media, and hypertargeted voter profiles.268 The capacity for distortion is not yet realized.

In 2010, amid growing concerns over consumer privacy online, the Digital Advertising Alliance launched the AdChoices program as part of the advertising industry’s self-regulation of online data tracking activities. A hallmark of the program is a little “i” icon found on online advertisements served to consumers based on data profiles. When consumers click on the little “i”, they gain information about why they were served that particular ad and have the opportunity to make choices about their online experience related to data collection and use.

The AdChoices program has increased consumer awareness about data collection and tracking online.269 Moreover, use of the icon led to increased trust of the advertiser.270 Survey respondents indicated support for continued use of the AdChoices program and the now-familiar icon; seventy percent agreed that policymakers considering new regulations should incorporate and encourage the use of the AdChoices program.271 This program has been the industry-led effort in the hybrid policy approach to data privacy regulations, paired with statutes, rules, and regulations on the commercial collection, use, and sharing of consumer data.

This type of combined market and legal solution should also be deployed to protect and enhance the flow of information influencing voting decisions. Lawmakers should ensure that the laws and rules governing misleading or false information related to elections are inclusive of AIgenerated content. Partnerships with tech platforms should encourage clearer and more conspicuous identifiers for AI-generated or manipulated content. Since some of the state laws require disclosures to appear in content metadata,272 tech platforms could search for this code and apply more user-friendly labels to AI content.

Enforcement of prohibitions on the use of AI or the dissemination of false or misleading information may present challenges and be delayed well beyond the conclusion of an election, leaving the harm to the voter without remedy. A hybrid approach that incorporates industry solutions, in

addition to legal ones, would more effectively address the role AI is playing in the erosion of trust in democracy and fair elections. Like the AdChoices program has raised awareness of the type of data collection advertisers engage in online, an AI content label could raise awareness among voters of the kind of information they are seeing populate their social media and news feeds. This awareness would be a real-time benefit that could curb the potential negative effect of misinformation. Platforms taking part in this program may also experience increased trust from their users if they participate in a transparent way—that is, publishing policies on how this content is identified and what choices users have to engage with the content and the label.

Like many forms of new media that have come before, artificial intelligence raises familiar questions with a novel framework. The increasing use and sophistication of AI will continue to challenge campaigns, lawmakers, and voters in the coming elections. Current campaign finance laws do little to counteract the possible negative effects of AI-generated content dominating political discourse. Newly proposed and adopted laws at the federal and state levels chip away at corners of the problem. A hybrid approach that encourages content creators and platforms to self regulate and engage in user awareness campaigns will provide a more balanced model to regulate election communications to preserve both election integrity and political speech freedoms.

--- FOOTNOTES ---

Part of the Election Law Commons, and the Legal Writing and Research Commons

* Courtney A. Barclay, Ph.D., J.D., is the Associate Dean for Academic and Faculty Affairs and a Professor of Law and Communication at Jacksonville University College of Law.

INTRODUCTION

1. Marshall Cohen, Fake Celebrity Endorsements Become Latest Weapon in Misinformation Wars, Sowing Confusion Ahead of 2024 Election, CNN (Aug. 22, 2024), https://www.cnn.com/2024/08/22/media/fake-celebrity-endorsements-social-media-2024election-misinformation/index.html [https://perma.cc/WKX5-MYLC]; Ellie Barnes, Top Takeaways: 3 Tips For Teaching News Literacy In The Age Of AI, NEWS LITERARY PROJ. (Oct. https://newslit.org/updates-and-events/topic/misinformation/ [https://perma.cc/8UC2-XXWU].

2024),

7,

2. Shannon Bond, How AI Deepfakes Polluted Elections in 2024, NPR: ALL THINGS 5:00 AM), https://www.npr.org/2024/12/21/nx-s1-

CONSIDERED 5220301/deepfakes-memes-artificial-intelligence-elections [https://perma.cc/4KCK-92C8].

(Dec.

21,

Id.

3. 4. Mike Vorhaus, People Increasingly Turn to Social Media for News, FORBES (June 24, 2020, 9:51 PM), https://www.forbes.com/sites/mikevorhaus/2020/06/24/peopleincreasingly-turn-to-social-media-for-news/ [https://perma.cc/LJJ2-9CCG].

Id.

5. 6. Shanay Gracia, Americans in Both Parties Are Concerned Over the Impact of AI (Sept. 19, 2024), the 2024 Presidential Campaign, PEW RSCH. CTR.

on

https://www.pewresearch.org/short-reads/2024/09/19/concern-over-the-impact-of-ai-on2024-presidential-campaign/ [https://perma.cc/N8DA-LMJ5].

7. Christopher St. Aubin & Jacob Liedke, Most Americans Favor Restrictions on False Information, Violent Content Online, PEW RSCH. CTR. (July 20, 2023), https://www.pewresearch.org/short-reads/2023/07/20/most-americans-favor-restrictions-onfalse-information-violent-content-online/ [https://perma.cc/8M3S-943R].

8. See ALEXANDER MEIKLEJOHN, FREE SPEECH AND ITS RELATION TO SELFGOVERNMENT 82–89 (1948); Thomas I. Emerson, Toward a General Theory of the First Amendment, YALE L. J. 877, 883–84 (1962–1963); see also e.g., Tinker v. Des Moines Indep. Cmty. Sch. Dist., 393 U.S. 503, 512 (1969); Healy v. James, 408 U.S. 169, 180–81 (1972).

9. Abrams v. United States, 250 U.S. 616, 630 (1919) (Holmes, J., dissenting). The marketplace of ideas theory refers to the work of John Stuart Mill, applying the economic theory of laissez-faire economics to political discourse. See, e.g., JOHN STUART MILL, ON LIBERTY 33–34 (3rd ed. 1864).

10. Brandenburg v. Ohio, 395 U.S. 444, 448–49 (1969). 11. See id. at 447–48. 12. See Stanley Ingber, The Marketplace of Ideas: A Legitimizing Myth, 33 DUKE L.J.

1, 36 (1984).

13. See Jerome A. Barron, Access to the Press—A New First Amendment Right, 80

HARV. L. REV. 1641, 1650–53 (1967).

14. Ingber, supra note 12, at 5. 15. See Theresa Josephine Seipp, Natali Helberger, Claes de Vreese & Jef Ausloos, Between the Cracks: Blind Spots in Regulating Media Concentration and Platform Dependence in the EU, 13 INTERNET POL’Y REV. 1, (2024).

16. David Gilbert, How Influencers and Algorithms are Creating Bespoke Realities for Everyone, WIRED (July 4, 2024), https://www.wired.com/story/influencers-algorithmsbespoke-reality/ [https://perma.cc/V36A-GLJE].

17. See, e.g., Buckley v. Valeo, 424 U.S. 1, 6 (1976); McConnell v. FEC, 124 S. Ct.

619, 656 (2003); Citizens United v. FEC, 130 S. Ct. 876, 886 (2010).

18. Federal Election Campaign Act of 1971, Pub. L. No. 92-225, 86 Stat. 3 (1972). 19. Sam Garrett, Artificial Intelligence (AI) and Campaign Finance Policy: Recent 2024),

Developments, https://crsreports.congress.gov/product/pdf/IN/IN12222 [https://perma.cc/6AY8-JNYB].

CONG.

SERV.

(Sept.

RSCH.

25,

20. See AI in Elections: A Look at the Federal and State Legislative Landscape, NAT’L CONF. OF STATE LEGISLATORS (Sept. 12, 2024), https://www.ncsl.org/elections-and-

campaigns/artificial-intelligence-ai-in-elections-and-campaigns LV3Y].

[https://perma.cc/2KN3-

21. Adav Noti, How Artificial Intelligence Influences Elections, and What We Can Do About It, CAMPAIGN LEGAL CTR. (Feb. 28, 2024), https://campaignlegal.org/update/howartificial-intelligence-influences-elections-and-what-we-can-do-about-it [https://perma.cc/MS8P-NVEW].

22. Id. 23. See, e.g., Curt Lyngaas, Donie Devine & Sean O’Sullivan, A Fake Recording of a Candidate Saying He’d Rigged the Election Went Viral. Experts Say It’s Only the Beginning, CNN. (Feb. 1, 2024), https://www.cnn.com/2024/02/01/politics/election-deepfake-threatsinvs/index.html [https://perma.cc/PY2F-RRRR]; Garance Burke, Chatbots’ Inaccurate, Misleading Responses About US Elections Threaten to Keep Voters from Polls, ASSOCIATED PRESS (Feb. 27, 2024, 4:06 PM), https://apnews.com/article/ai-chatbots-elections-artificialintelligence-chatgpt-falsehoods-cc50dd0f3f4e7cc322c7235220fc4c69 [https://perma.cc/TCX4-VPS7]. 24. Burke, supra note 23. 25. See Fraudulent Misrepresentation of Campaign Authority, 89 FED. REG. 78785

(Sept. 26, 2024).

26. Artificial Intelligence (AI) in Elections and Campaigns, NAT’L CONF. OF ST. LEGISLATURES (Oct. 10, 2023), https://www.ncsl.org/elections-and-campaigns/artificialintelligence-ai-in-elections-and-campaigns [https://perma.cc/YQY5-ZTBB].

27. Daniel Weiner & Lawrence Norden, Regulating AI Deepfakes and Synthetic the Political Arena, BRENNAN CTR. FOR JUST. 1, 1 (Dec. 5, 2023),

Media https://www.brennancenter.org/our-work/research-reports/regulating-ai-deepfakes-andsynthetic-media-political-arena [https://perma.cc/A5ZH-YQZW].

CAMPAIGNS IN THE AGE OF AI

28. One Big Beautiful Bill Act, H.R. 1, 119th Cong. § 43201(c) (May 20, 2025). 29. Supreme Court Campaign Finance Decision Favors an Open Marketplace of Ideas, FIRE (Jan. 21, 2010), https://www.thefire.org/news/supreme-court-campaign-financedecision-favors-open-marketplace-ideas [https://perma.cc/SR7Y-WRY7].

30. Rick Spair, The Future of Social Media: How Generative AI is Revolutionizing the Way We Connect, MEDIUM (Jan. 19, 2024), https://medium.com/@rickspair/the-future-ofsocial-media-how-generative-ai-is-revolutionizing-the-way-we-connect-3df7a5489473 [https://perma.cc/JZL4-4ULB].

31. Bernard Marr, 15 Amazing Real-World Applications of AI Everyone Should Know About, FORBES (May 10, 2023, 2:51 AM), https://www.forbes.com/sites/bernardmarr/ 2023/05/10/15-amazing-real-world-applications-of-ai-everyone-should-know-about/ [https://perma.cc/9SSE-HJ4L].

32. See, e.g., Andrew R. Chow, AI’s Underwhelming Impact on the 2024 Elections, https://time.com/7131271/ai-2024-elections/

7:00 AM),

(Oct.

TIME [https://perma.cc/PHB4-B3V4].

30,

33. Memorandum from Ellen L. Weintraub, Chair, Fed. Election Comm’n, to Comm’n Sec’y (June 13, 2019) (on file with Fed. Election Comm’n); 2024 Political Ad Spending Will Jump Nearly 30% vs. 2020, EMARKETER (Jan. 11, 2024), https://www.emarketer.com/pressreleases/2024-political-ad-spending-will-jump-nearly-30-vs-2020/ [https://perma.cc/VU96LVU6].

34. Sheera Frankel, The Year of the A.I. Election That Wasn’t, N.Y. TIMES (Aug. 21,

2024), https://www.nytimes.com/2024/08/21/technology/ai-election-campaigns.html [https://perma.cc/A6F3-X9UH]. There are some exceptions to this, including Former President Trump’s campaign using AI-generated content to fake an endorsement from Taylor Swift and share memes related to immigrants eating cats, although voters and commentators alike quickly identified these as AI-generated images. Id.

35. Shanay Gracia, Americans in Both Parties Are Concerned Over the Impact of AI

on the 2024 Presidential Campaign, PEW RSCH. CTR. (Sept. 19, 2024), https://www.pewresearch.org/short-reads/2024/09/19/concern-over-the-impact-of-ai-on2024-presidential-campaign/ [https://perma.cc/N8PM-K9LR].

36. E.g., Jared Schroeder, Saving the Marketplace from Market Failure: Reorienting

Marketplace Theory in the Era of AI Communicators, 28 WM. & MARY BILL RTS. J. 689, 696 (2020).

37. Laurie A. Harris, Artificial Intelligence: Overview, Recent Advances, and

Considerations for the 118th Congress, CONG. RSCH. SERV.(Aug. 4, 2023), https://crsreports.congress.gov/product/pdf/R/R47644 [https://perma.cc/9ELA-FRQ3].

38. Id. 39. Id.

40. Id. 41. Id. 42. Id. 43. Weiner & Norden, supra note 27. 44. Oversight of A.I.: Rules for Artificial Intelligence: Hearing on the Oversight of Artificial Intelligence Before the Subcomm. on Priv., Tech,. & the Law of the S. Comm. on the Judiciary, 118th Cong. 14–15 (May 23, 2023) (statement by Sam Altman, CEO, OpenAI).

45. Bruce Schneier, The Coming AI Hackers, BELFER CTR. FOR SCI. & INT’L AFFS., https://www.schneier.com/wp-

HARV. content/uploads/2021/04/The-Coming-AI-Hackers.pdf [https://perma.cc/5BWK-WNGF]. 46. Andrew Myers, AI’s Powers of Political Persuasion, HAI STANFORD UNIV.

KENNEDY

2021),

(Apr.

SCH.

(Feb. [https://perma.cc/K8RG-CASG].

2023),

27,

https://hai.stanford.edu/news/ais-powers-political-persuasion

47. Id. 48. See Hui Bai et al., Artificial Intelligence Can Persuade Humans on Political Issues, OSF PREPRINTS 1, 10–11, https://osf.io/preprints/osf/stakv [https://perma.cc/DXP8898N] (last visited Oct. 2023).

49. Id.; Max Weiss, Deepfake Bot Submissions to Federal Public Comment Websites Cannot Be Distinguished from Human Submissions, TECH. SCI. 1, 2 (Dec. 17, 2019), https://techscience.org/a/2019121801/ [https://perma.cc/382B-UBZ6]. Weiss notes a 2017 study of comments submitted to the Federal Communications Commission during a public commenting period, the majority of which were later determined to be submitted by chatbots.

50. Id. 51. Id. 52. Schneier, supra note 45, at 14. 53. See, e.g., Jason Murdock, What is the Internet Research Agency? Facebook Shuts Hundreds of Accounts Linked to Russian Troll Factory, NEWSWEEK (Apr. 4, 2018, 5:29 AM), https://www.newsweek.com/what-internet-research-agency-facebook-shuts-hundredsaccounts-linked-russia-870889 [https://perma.cc/LL8R-UZQQ].

54. Emily Dreyfuss, Alert: Don't Believe Everything You Read About the Migrant Caravan, WIRED (Oct. 23, 2018), https://www.wired.com/story/mexico-migrant-caravanmisinformation-alert/ [https://perma.cc/5XGB-LUHX ].

55. Jeff Berkowitz, The Evolving Role of Artificial Intelligence and Machine Learning (Dec. 21, 2020),

in US Politics, CTR. https://www.csis.org/blogs/technology-policy-blog/evolving-role-artificial-intelligence-andmachine-learning-us-politics [https://perma.cc/NBK2-CEG3].

FOR STRATEGIC &

INT'L STUDS.

56. Risk in Focus: Generative A.I. and the 2024 Election Cycle, CYBERSEC & 2024),

AGENCY

SEC.

18,

INFRASTRUCTURE https://www.cisa.gov/sites/default/files/202405/Consolidated_Risk_in_Focus_Gen_AI_ElectionsV2_508c.pdf N8JB].

(Jan.

[https://perma.cc/4X7Y-

57. Id.

THE MARKETPLACE OF IDEAS

58. Id. 59. Zelly Martin, et al., Political Machines: Understanding the Role of AI in the U.S. 2024 Elections and Beyond, CTR. FOR MEDIA ENGAGEMENT 1, 7 (June 6, 2024), https://mediaengagement.org/research/generative-ai-elections-and-beyond [https://perma.cc/N5L9-SM9D].

60. Id. at 7–8. 61. See Chow, supra 32. 62. See Abrams v. U.S., 250 U.S. 616, 630 (1919) (Holmes, J., dissenting). 63. Id. at 618–19. 64. Id. at 630. 65. Lamont v. Postmaster Gen. of the U.S., 381 U.S. 301, 308 (1965) (Brennan, J.,

concurring).

66. Id. 67. Id. at 308. 68. Id. 69. Keyishian v. Bd. of Regents of the Univ. of the State of New York, 385 U.S. 589,

603 (1967).

70. See, e.g., Tinker v. Des Moine Indep. Cmty. Sch. Dist., 393 U.S. 503, 512 (1969);

Healy v. James, 408 U.S. 169, 180–81 (1972).

71. See, e.g., Barron, supra note 13, at 1641; C. Edwin Baker, Scope of the First

Amendment Freedom of Speech, 25 UCLA L. REV. 964, 974–75 (1978).

72. Barron, supra note 13. 73. Id. at 1642–43. 74. Id. 75. E.g., id. at 1647–48. 76. Id. at 1656. 77. Id. at 1660–61. 78. Id. at 1656.

79. Id. 80. Id. 81. Id. at 1666–68. 82. Id. at 1647. 83. Id. at 1678. 84. Baker, supra note 71. 85. Id. at 974. 86. Id. at 976. 87. Id. at 984. 88. Id. at 967. 89. Id. at 974–81. 90. Id.at 981–90. 91. Id. 92. Id.at 989–90.

CAMPAIGN FINANCE REFORM AS A MARKET CORRECTOR

93. Id.at 990. 94. Red Lion Broad. Co. v. F.C.C., 395 U.S. 367, 389–90 (1969). 95. Kleindienst v. Mandel, 408 U.S. 753, 763 (1972); Miami Herald Publ’g Co. v. Tornillo, 418 U.S. 241, 251 (1974); Bigelow v. Virginia, 421 U.S. 809, 826 (1975); Va. State Bd. of Pharmacy v. Va. Citizens Consumer Council, 425 U.S. 728, 760 (1976).

96. See Bradley A. Smith, The Sirens’ Song: Campaign Finance Regulation and the First Amendment, 6 J.L. & POL’Y 1 1, 9 (1997). Fundraising became an important aspect of a campaign. Corporate financing significantly contributed to the election of William McKinley as the country’s 25th president over William Jennings Bryan. Id. at 10.

97. Id. 98. Cost of Election, OPEN SECRETS, https://www.opensecrets.org/electionsoverview/cost-of-election [https://perma.cc/MJ25-PUG3]; Fundraising Totals: Who Raised the Most?, OPEN SECRETS, https://www.opensecrets.org/elections-overview/fundraisingtotals [https://perma.cc/QQR6-LFVW].

99. See Smith, supra note 96, at 10, 18. 100. See ROBERT E. MUTCH, CAMPAIGNS, CONGRESS, AND COURTS: THE MAKING OF FEDERAL CAMPAIGN FINANCE LAW at xvii (1988). By the early 1930s, thirty-five states had passed legal prohibitions or limitations on direct contributions by corporations. See LOUISE

OVERACKER, MONEY IN ELECTIONS, POLITICS AND PEOPLE: THE ORDEAL OF SELFGOVERNMENT IN AMERICA 293–94 (1932).

101. Tillman Act, ch. 420, 34 Stat. 864–65 (1907); see also MUTCH, supra note 100,

at 6.

102. See Federal Election Campaign Act of 52 U.S.C. §§ 30101–30146. The law defines contribution as “any gift, subscription, loan, advance, or deposit of money or anything of value made by any person for the purpose of influencing any election for Federal office; or the payment by any person of compensation for the personal services of another person which are rendered to a political committee without charge for any purpose.” Id. § 30101 (8)(A). Independent expenditures are “an expenditure by a person—(A) expressly advocating the election or defeat of a clearly identified candidate; and (B) that is not made in concert or cooperation with or at the request or suggestion of such candidate, the candidate’s authorized political committee, or their agents, or a political party committee or its agents.” Id. § 30101 (17).

103. See id. § 30116; see also Contribution Limits, FED. ELECTION COMM’N,

https://www.fec.gov/help-candidates-and-committees/candidate-takingreceipts/contribution-limits/ [https://perma.cc/6YGC-N4JU] (last visited Dec. 12, 2024).

104. See 52 U.S.C. § 30118 (corporations and labor unions); id. § 30121 (foreign nationals); see also Who Can and Can’t Contribute, FED. ELECTION COMM’N, https://www.fec.gov/help-candidates-and-committees/candidate-taking-receipts/who-canand-cant-contribute/ [https://perma.cc/W7C9-ZC5B] (last visited Dec. 12, 2024).

105. See Candidate Contributions, FED. ELECTION COMM’N, https://www.fec.gov/helpcandidates-and-committees/filing-reports/candidate-contributions/ [https://perma.cc/R3M7LPFK] (last visited Dec. 12, 2024); PACs & Other Political Committee Contributions, FED. https://www.fec.gov/help-candidates-and-committees/filingELECTION reports/pacs-other-political-committee-contributions/ (last visited Dec. 12, 2024); Individual Contributions, FED. ELECTION COMM’N, https://www.fec.gov/help-candidates-and-committees/filing-reports/individual-contributions/ [https://perma.cc/2VKH-KEXQ] (last visited Dec. 12, 2024).

[https://perma.cc/P9WN-7BK]

COMM’N,

106. See

Advertising

and

Disclaimers,

FED.

ELECTION

COMM’N,

https://www.fec.gov/help-candidates-and-committees/advertising-and-disclaimers/ [https://perma.cc/TMZ5-RKK6] (last visited Dec. 12, 2024).

107. See Smith, supra note 96, at 4–5. 108. See Smith, supra note 96, at 1, 7. 109. See e.g., Bipartisan Campaign Reform Act of 2002, Pub. L. No. 107-155, 116 Stat. 81 (codified in scattered sections of 2 U.S.C. and later transferred to scattered sections of 52 U.S.C.) (tightening the FECA disclosure requirements); Citizens United v. Fed. Election Comm’n, 130 S. Ct. 876 (2010) (upholding disclaimer and disclosure provisions).

110. See Citizens United, 130 S. Ct. at 911 (“The remedies enacted by law, however, must comply with the First Amendment; and, it is our law and tradition that more speech, not less, is the governing rule.”); see also David L. Hudson, Counterspeech Doctrine, FREE SPEECH CTR. (July 2, 2024), https://firstamendment.mtsu.edu/article/counterspeech-doctrine/ [https://perma.cc/H3YU-DYX2].

111. But see infra Section III.A. 112. See Justin A. Nelson, The Supply and Demand of Campaign Finance Reform, 100

COLUM. L. REV. 524, 534 (2000).

113. Tillman Act, ch. 420, 34 Stat. 864–65 (1907). 114. Nelson, supra note 112, at 535–37. 115. Federal Election Campaign Act of 52 U.S.C. §§ 30101–30146.

116. See id. § 30118 (contributions and expenditures by corporations); id. § 30125

(candidates).

117. Laura Christine Dunn, Effects of the Bipartisan Campaign Finance Reform Act on Federal Congressional Candidates: A Case Study (Aug. 2005) (M.A. thesis, University of Florida).

118. Federal Election Campaign Act Amendments of 1974, Pub. L. No. 93-443, 88 Stat. 1280 (current version at 52 U.S.C. § 30106). Individuals’ contributions were limited to $1,000 per candidate and $25,000 in total contributions; expenditures were limited to $1,000 per candidate per year. Federal Election Campaign Act Amendments of 1974, Pub. L. 93443, 88 Stat. 1263.

119. Federal Election Campaign Act Amendments of 1974, Pub. L. 93-443, 88 Stat. 1263. In election years, campaign committees and active political committees were required to file a financial report each quarter. If a committee received or spent $1,000 or more in the quarter, then additional reports had to be filed ten days before and thirty days after each election. Contributions of more than $1,000 that were received within fifteen days of an election had to be reported within 48 hours. In non-election years, committees were required to file year-end reports.

120. Buckley v. Valeo, (1976). 121. See generally id. 122. See generally id. 123. See id. at 51–52. 124. Id. at 57–58.

125. Id. at 66–68. 126. Id. at 68. The Court also addressed whether this balance changed when disclosure requirements were applied to minor parties or independents. The Court found that there may be the possibility that disclosure would be too burdensome, but that a general exemption was too broad and refused to create one. This area of disclosure law is extremely interesting, but outside the scope of this paper.

127. See Federal Election Campaign Act Amendments of 1976, Pub. L. No. 94-283, 90

Stat. 487–88 (current version at 52 U.S.C. § 30116).

128. See Federal Election Campaign Act Amendments of 1976, Pub. L. No. 94-283, 90 Stat. 480 (current version at 52 U.S.C. § 30102). This included mandatory reporting of expenditures and a sworn statement as to whether the expenditure was “made in cooperation, consultation, or concert with, or at the request or suggestion of, any candidate or any authorized committee or agent of such candidate.” Id.

129. Bipartisan Campaign Reform Act of 2002, Pub. L. No. 107-155, 116 Stat. 81 (codified in scattered sections of 2 U.S.C. and later transferred to scattered sections 52 U.S.C.), invalided in part by Citizens United v. Fed. Election Comm’n, 558 U.S. 310 (2010). On February 13, 2002, the House of Representatives passed H.R. 2356. The bill was then adopted by the Senate on March 18 and 20, 2002. President George W. Bush signed H.R. 2356 into law on March 27, 2002. BCRA is also known as the McCain-Feingold Act, after its Senate co-sponsors, Senators John McCain and Russell Feingold.

130. See 148 Cong. Rec. S2104–S2105 (daily ed. Mar. 20, 2002) (statement of Sen. Feingold). These funds were generally raised for grass-roots campaigns and party-specific activities (i.e., bumper stickers, get-out-the-vote drives, and committee publications with three or more candidates listed). For a further discussion of soft money, see FED. ELECTION COMM’N, TWENTY YEAR REPORT 23 (1995), https://www.fec.gov/resources/cmscontent/documents/20yearreport.pdf [https://perma.cc/6PWA-3W8P].

131. See ANTHONY CORRADO, MONEY AND POLITICS: A HISTORY OF FEDERAL

CAMPAIGN FINANCE LAW CAMPAIGN FINANCE REFORM: A SOURCEBOOK (1997).

132. Bipartisan Campaign Reform Act of 2002, § 307(a)(1), Pub. L. No. 107-155, 116 Stat. 102 (current version at 52 U.S.C. § 30118); see also Bipartisan Campaign Reform Act of 2002, § 203, Pub. L. 107-155, 116 Stat 91–92 (codified in scattered sections of 52 U.S.C.).

133. See Bipartisan Campaign Reform Act of 2002, Pub. L. No. 107–155 (codified in

scattered section of 52 U.S.C.).

134. Id. § 301(22) (current version at 52 U.S.C. § 30101). 135. Id. § 301(20) (current version at 52 U.S.C. § 30101). 136. Id. § 301(20)(iii) (current version at 52 U.S.C. § 30101). 137. Id. § 323 (current version at 52 U.S.C. § 30125). 138. 11 CFR § 110.11 (2024). 139. TREVOR POTTER, The Current State of Campaign Finance Law, in THE NEW

CAMPAIGN FINANCE SOURCEBOOK 48, 56 (Anthony Corrado et al. eds., 2005).

140. Bipartisan Campaign Reform Act of 2002, § 203, Pub. L. No. 107-55, 116 Stat.

81, 91–92 (current version at 52 U.S.C. § 30118).

141. Id. An incorporated state party or candidate committee may make these communications if it is not a political committee, is incorporated for liability purposes only, does fund electioneering communications with corporate or labor union donations, and meets all reporting requirements.

142. Id. 143. See id. § 203, invalidated by Citizens United v. Fed. Election Comm’n, 558 U.S.

310 (2010).

144. Citizens United, 558 U.S. at 360. 145. See McConnell v. Fed. Election Comm’n, 124 S. Ct. 619 (2003). 146. Id. at 635. 147. Citizens United, 558 U.S. at 360. 148. McCutcheon v. Fed. Election Comm’n, 572 U.S. 185, 227 (2014). 149. See id. at 224.

150. See id. 151. Id. 152. See id. 153. Id. 154. Id. 155. Id. 156. Citizens United v. Fed. Election Comm’n, , 371 (2010). 157. See Artificial Intelligence (AI) in Elections and Campaigns, NAT’L CONF. OF https://www.ncsl.org/elections-and2024), [https://perma.cc/S86Z-

(Oct. STATE campaigns/artificial-intelligence-ai-in-elections-and-campaigns G5G8].

LEGISLATURES

24,

158. Lear Jiang, Disclosure’s Last Stand? The Need to Clarify the “Informational

Interest” Advanced by Campaign Finance Disclosure, 119 COLUM. L.R. 487, 495 (2019).

159. Taylor Giorno, Opaque Nonprofit Donation to Shadowy Super PAC Raises a.m.),

‘Several https://thehill.com/homenews/campaign/5034983-opaque-nonprofit-donation-to-shadowysuper-pac-raises-several-red-flags/ [https://perma.cc/3TYC-Q527].

Flags’,

(Dec.

HILL

6:00

Red

12,

160. Campaign Finance Law: An Analysis of Key Issues, Recent Developments, and for Legislation, CRS REPORTS, May 17, 2023,

Constitutional Considerations https://crsreports.congress.gov/product/pdf/R/R45320?utm_source=chatgpt.com [https://perma.cc/2NMH-SGDJ].

161. Luke Wachob, Misusing Disclosure: How a Policy Intended to Increase Voter Knowledge Often Misleads the Public, INST. FOR FREE SPEECH (Aug. 18, 2014), https://www.ifs.org/research/misusing-disclosure-how-a-policy-intended-to-increase-voterknowledge-often-misleads-the-public/?utm_source=chatgpt.com PBKD].

[https://perma.cc/4UGT-

162. See Abby K. Wood, Learning from Campaign Finance Information, 70 EMORY L. J. https://scholarlycommons.law.emory.edu/elj/vol70/iss5/2 [https://perma.cc/4LUP-WS58]; Jennifer A. Heerwig, Diagnosing Disclosure: A Social Scientific Perspective on Campaign Finance Disclosure, YALE L. & POL’Y REV. 8, 9 (Apr. 3, 2016).

(2021),

163. Zachary Newkirk, From Slaves to Pornographers, Campaign Contribution (Mar. 31, 2011, 11:37 a.m.),

Records Rife With Characters, OPEN SECRETS https://www.opensecrets.org/news/2011/03/from-slaves-to-pornographers-campaigncontribution-records/ [https://perma.cc/G24E-2ESD].

164. Buckley v. Valeo, , 14 (1976). For a full discussion of

the Buckley decision, see infra Section III.

165. Buckley, 424 U.S. at 64. 166. See, e.g., id.; Citizens Against Rent Control/Coalition for Fair Hous. v. Berkeley, 454 U.S. 290 (1981); Fed. Election Comm’n. v. Mass. Citizens for Life, Inc., 479 U.S. 238 (1986); Austin v. Mich. Chamber of Com. 494 U.S. 652 (1990); McIntyre v. Ohio Elections Comm’n, 514 U.S. 334 (1995); Citizens United v. Fed. Election Comm’n, 130 S. Ct. 876, 910 (2010).

167. See, e.g., Mass. Citizens for Life, Inc., 479 U.S. at 257; Austin, 494 U.S. at 652;

McConnell v. Fed. Election Comm’n, 124 S. Ct. 619, 693 (2003).

168. Cass R. Sunstein, Exchanges in the Welfare State: Free Speech Now, 59 UNIV.

CHI. L. REV. 255, 291 (1992).

169. Id. at 291 (quoting at 48–49). 170. Kathleen M. Sullivan, Political Money and Freedom of Speech, 30 U.C. DAVIS L.

REV. 663, 671–86 (1997).

171. FEC v. Nat’l Conservative PAC, 470 U.S. 480, 496–97 (1985). 172. See id. at 496. 173. See Thomas F. Burke, The Concept of Corruption in Campaign Finance Law,

CONST. COMMENT. 127, 128–30 (1997).

174. John G. Peters & Susan Welch, Political Corruption in America: A Search for Definitions and a Theory, or If Political Corruption Is in the Mainstream of American Politics Why Is It Not in the Mainstream of American Politics Research?, 72 AM. POL. SCI. REV. POLITICAL 974, CORRUPTION (1972 .

(citing JAMES SCOTT,

974–75 (1978)

COMPARATIVE

175. Id. at 974–75 (citing LARRY L. BERG ET AL., CORRUPTION IN THE AMERICAN

POLITICAL SYSTEM (1976).

176. Id. at 975 (quoting ARNOLD A. ROGOW & HAROLD LASSWELL, POWER CORRUPTION

AND RECTITUDE 132–33 (1963 .

177. Id. The authors provide the example of a politician ‘fixing’ immigration papers for

an illegal alien who has aided economic growth.

178. Id. 179. Id. 180. Id. 181. Id. at 976. 182. Id. at 976–78. The authors then tested the design with ten different scenarios: 1) A presidential candidate who promises an ambassadorship in exchange for campaign contributions; 2) a member of Congress using seniority to obtain a weapons contract for a firm in his or her district; 3) a public official using public funds for personal travel; 4) a secretary of defense who owns $50,000 in stock in a company with which the Defense Department has a million-dollar contract; 5) a public official using influence to get a friend or relative admitted to law school; 6) the driveway of the mayor's home being paved by the city crew; 7) a state assembly member while chairperson of the public roads committee authorizing the purchase of land s/he had recently acquired; 8) a judge with $50,000 worth of stock in a corporation hearing a case concerning that firm; 9) a legislator accepting a large campaign contribution in return for voting "the right way" on a legislative bill; 10) a member of Congress who holds a large amount of stock (about $50,000 worth) in Standard Oil of New Jersey working to maintain the oil depletion allowance. Id. at 978.

183. Sullivan, supra note 170, at 679. 184. Id. 185. Id. 186. E.g., id.

187. Id. at 679–80. 188. Id. at 680. 189. Id. 190. See Robert F. Bauer, The Demise of Reform: Buckley v. Valeo, The Courts, and the

“Corruption Rationale”, 10 STAN. L. & POL’Y REV. 11, 11 (1998).

191. Id. at 12–13. 192. See Fed. Election Comm’n v. Nat’l Conservative Pol. Action Comm., 470 U.S.

480 (1985).

193. See id. 194. Id. at 497. 195. Bauer, supra note 190, at 14. 196. Burke, supra note 173, at 131. 197. Id.; see also Buckley v. Valeo, , 36 (1976). 198. Buckley, 424 U.S. at 26. But see id. at 28 (arguing that laws focusing on the prevention of bribery alone “deal with only the most blatant and specific attempts of those with money to influence governmental action”). 199. Burke, supra note 173, at 131.

200. Fed. Election Comm’n v. Nat’l Conservative Pol. Action Comm., 470 U.S. 480,

497 (1985).

201. Id. (quoting Buckley, 424 U.S. at 47). 202. Austin v. Mich. Chamber of Com., 494 U.S. 652, 659–60 (1990), overruled by

Citizens United v. Fed. Election Comm’n, (2010).

203. Id. 204. Id. at 685 (Scalia, J., dissenting). 205. See David Strauss, Corruption, Equality, and Campaign Finance Reform, 94

COLUM. L. REV. 1369, 1370 (1994).

206. See id. 207. Id. at 1369. 208. Buckley v. Valeo, , 26 (1976).

209. Sullivan, supra note 170, at 672. 210. Id. 211. Id. at 673. 212. Brief for the Attorney General and the FEC at 23, Buckley v. Valeo,

(1976) (per curiam) (Nos. 75-436, 75-437), 1975 WL 171459 [hereinafter Buckley brief].

213. Id. 214. Buckley v. Valeo, , 26 (1976). 215. Id. at 48–49. 216. Id.

217. Id. at 265 (White, J., concurring in part and dissenting in part). 218. JOHN RAWLS, POLITICAL LIBERALISM 360–61 (expanded ed. 2005). 219. John McCain, Reclaiming Our Democracy: The Way Forward, 3 ELECTION L.J.

115, 115 (2004).

220. Id. at 120–21. 221. Landell v. Sorrell, 406 F.3d 159, 162 (2d Cir.) (Calabresi, J., concurring), cert. granted sub nom., Randall v. Sorrell, 548 U.S. 230, (2006); see also Jacobus v. Alaska, 338 F.3d 1095, 1107 (9th Cir. 2003) (upholding limits on soft money contributions in Alaska), overruled by Bd. of Trustees of Glazing Health & Welfare Tr. V. Chambers, 941 F.3d 1195 (9th Cir. 2019). In Jacubus, the 9th Circuit panel held that unregulated campaign finance creates an environment in which individual citizens’ voices are “drown[ed] out” by wealth.” Id.

222. Cass R. Sunstein, Political Equality and Unintended Consequences, 94 COLUM. L.

REV. 1390, 1400 (1994).

223. Richard L. Hasen, Campaign Finance Laws and the Rupert Murdoch Problem, 77 TEX. L. REV. 1627, 1629 (1999) (discussing the media exception built into the campaign finance reform structure).

224. Jamin Raskin & John Bonifaz, The Constitutional Imperative and Practical

Superiority of Democratically Financed Elections, 94 COLUM. L. REV. 1160, 1165 (1994).

225. See generally BRADLEY A. SMITH, UNFREE SPEECH: THE FOLLY OF CAMPAIGN

FINANCE REFORM (2003).

226. Richard L. Hasen, Buckely is Dead, Long Live Buckley: The New Campaign Finance Incoherence of McConnell v. Federal Election Commission, 153 U. PA. L. REV. 31, 31–32 (2004).

227. Id. at 57. 228. Nixon v. Shrink Mo. Gov’t PAC, 528 U.S. 377, 402 (2000) (Breyer, J.,

concurring). 229. Id. 230. Id.

AI DISCLOSURES AND PROHIBITIONS

231. See Hasen, supra note 226, at 57–58. 232. Id. 233. Citizens United v. Fed. Election Comm’n, , 440 (2009). 234. Schroeder, supra note 36, at 689.

235. See Tom Ramstack, Congress Wants Laws to Confront Deepfakes that Deceive 2023),

Consumers, https://www.thewellnews.com/technology/congress-wants-laws-to-confront-deepfakes-thatdeceive-consumers/ [https://perma.cc/T8Z8-YX2A].

NEWS

WELL

(Nov.

9,

236. Fraudulent Misrepresentation of Campaign Authority, 89 Fed. Reg. 78785 (Sept.

26, 2024) (to be codified at 11 C.F.R. pt. 110).

237. Commission approves Notification of Disposition, Interpretive Rule on Artificial (Sept. 27, 2024),

Intelligence https://www.fec.gov/updates/commission-approves-notification-of-disposition-interpretiverule-on-artificial-intelligence-in-campaign-ads/ [https://perma.cc/3AKR-WAKU].

in Campaign Ads, FED. ELECTION COMM’N

238. See Statement of Vice Chair Ellen L. Weintraub on the Disposition of the Rulemaking Petition Regarding Fraudulent Misrepresentation and Artificial Intelligence (Sept. 19, 2024), https://www.fec.gov/resources/cms-content/documents/REG-2023-02-Ain-Campaign-Ads-Vice-Chair-Statement.pdf [https://perma.cc/3EX3-ZQVS].

239. See Fraudulent Misrepresentation of Campaign Authority, 89 Fed. Reg. 78785

(Sept. 26, 2024) (to be codified at 11 C.F.R. pt. 110).

240. Id. 241. See Artificial Intelligence (AI) in Elections and Campaigns, NAT’L CONF. OF https://www.ncsl.org/elections-and2025), [https://perma.cc/S86Z-

(Apr. STATE campaigns/artificial-intelligence-ai-in-elections-and-campaigns G5G8].

LEGISLATURES

10,

242. See, e.g., ALA. CODE § 17-5-16.1 (2024). 243. See TEX. ELECTION CODE § 255.004 (2023); MINN. STAT. ANN. § 609.771 (2024). 244. See, e.g., ALA. CODE § 17-5-16.1 (2024). 245. See, e.g., TEX. ELECTION CODE § 255.004 (2023); MICH. COMP. LAWS. § 168.932f (2024) (“materially deceptive media”); N.H. REV. STAT. ANN. § 664:14-c (2024) (“synthetic media”).

246. See Artificial Intelligence (AI) in Elections and Campaigns, supra note 241. 247. See, e.g., ALA. CODE § 17-5-16.1 (2024) (type of disclaimer required); ARIZ. REV. STAT. ANN.§ 16-1023 (2024) (proximity of election required); MINN. STAT. ANN. § 609.771 (2024) (intent required).

248. See, e.g., ALA. CODE § 17-5-16.1 (2024). 249. See, ALA. CODE § 17-5-16.1 (2024); ARIZ. REV. STAT. ANN. § 16-1023 (2024); COLO. REV. STAT. § 1-46-103 (2024); DEL. CODE ANN. tit. 15, § 5145 (2024); MICH. COMP. LAWS. § 168.932f (2024); MINN. STAT. ANN. § 609.771 (2024); N.M. STAT. ANN. § 1-1926.8 (2024); N.H. REV. STAT. ANN. § 664:14-c (2024); TEX. ELECTION CODE § 255.004 (2023).

250. See CAL. ELECTION CODE § 20012 (2024). 251. See FLA. STAT. § 106.145 (2024); HAW. REV. STAT. § 11-303 (2024); IDAHO CODE § 67-6628 (2024); IND CODE § 3-9-8-5 (2024); N.Y. ELECTION LAW § 14-106 (2024); OR. REV. STAT. § 260.345; UTAH CODE ANN. § 20A-11-1104; WASH. REV. CODE § 29B.35.010 (2024); WIS. STAT. § 11.1303 (2024).

252. See Artificial Intelligence (AI) in Elections and Campaigns, supra note 241 (detailing in a chart which states provide civil remedies). These states include Arizona, California, Colorado, Hawaii, Idaho, Indiana, New Mexico, New Hampshire, New York, Oregon, Utah, Washington, and Wisconsin. Id.

253. See id. (detailing in a chart which states provide criminal remedies). These states

include Florida and Texas.

254. See id. (detailing in a chart which states provide both civil and criminal remedies).

These states include Alabama, Delaware, Michigan, and Minnesota.

255. See AI in Elections: A Look at the Federal and State Legislative Landscape, NAT’L CONF. OF STATE LEGISLATURES (Sept. 12, 2024), https://www.ncsl.org/elections-andcampaigns/artificial-intelligence-ai-in-elections-and-campaigns [https://perma.cc/S86ZG5G8].

256. See id. 257. See Protect Elections from Deceptive AI Act , S. 2770, 118th Cong. (2024). 258. AI Transparency in Elections Act of 2024, S. 3875, 118th Cong. (2024). 259. Id. 260. See, e.g., HAW. REV. STAT. § 11-303 (2024). 261. See AI Transparency in Elections Act of 2024, S. 3875, 118th Cong. (2024). 262. Protect Elections from Deceptive AI Act , S. 2770, 118th Cong. (2024). 263. Id.

A HYBRID APPROACH TO AI POLICY

264. Id. 265. Id. 266. Kohls v. Bonta, Case No. 2:24-cv-02527-JAM-CKD, Order Granting Plaintiff’s Motion for Preliminary Injunction (E.D. Cal. Oct. 2, 2024); see Aaron Mak, A California Judge Opens (Aug. 7, 2025), https://www.politico.com/newsletters/digital-future-daily/2025/08/07/a-california-judgeopens-the-door-to-election-deepfakes-00498484 [https://perma.cc/J6L7-UETZ].

to Election Deepfakes, POLITICO

the Door

267. Gabriel R. Sanchez & Keesha Middlemass, Misinformation is Eroding the 2022),

Public’s https://www.brookings.edu/articles/misinformation-is-eroding-the-publics-confidence-indemocracy/ [https://perma.cc/SZS2-46X5].

Democracy,

BROOKINGS

(July.

Trust

26,

268. Barron, supra note 13, 1647–48. 269. Survey Finds Broad Consumer Recognition, Trust, Understanding, and Support for DAA’s AdChoices (Apr. 2, 2024), Icon, DIGITAL ADVERTISING ALLIANCE https://digitaladvertisingalliance.org/press-release/survey-finds-broad-consumer-recognitiontrust-understanding-and-support-daas-adchoices [https://perma.cc/JN62-58N6].

270. Id. 271. Id. 272. Sanam Hooshidary & Adam Kuckuk, The Use of Artificial Intelligence in Elections: The Federal and State Legislative Landscape, NAT’L CONF. OF STATE (Sept. 12, 2024), https://www.ncsl.org/elections-and-campaigns/ai-inLEGISLATURES elections-a-look-at-the-federal-and-state-legislative-landscape [https://perma.cc/75UKGMTJ].

CONCLUSION