*Chauncey Bellamy
I. Introduction
A shirtless, raptor-riding, pistol-brandishing potential future United States president has found a way to stand out among his fellow 2028 shadow presidential candidates.[1] In August 2025, Tucker Carlson, Kid Rock, and an angelic Hulk Hogan gave him their blessing during a solemn prayer session.[2] The next day, he stared sternly from his perch next to George Washington on Mount Rushmore.[3] This is the social-media reality of California Governor Gavin Newsom.[4] In a series of AI-generated X posts, Newsom and those inspired by him online have parodied President Trump’s social-media style in an attempt to redefine the role of the opposition.[5] At the same time, Newsom and other politicians across the country have struggled to limit a more insidious use of AI in the realm of politics: deepfakes designed to trick voters into turning away from political candidates for saying or doing things the candidates did not say or do.[6]
II. A Tale of Two California AI Laws
In 2024, California “enacted two political deepfake laws in the lead-up to the presidential election.”[7] One law “required social media platforms to remove or label ‘materially deceptive’ AI-generated political content near elections.”[8] Senior United States District Judge John A. Mendez, however, struck down the law because the federal Communications Decency Act protects internet sites from liability for posts by third parties.[9] The other law banned any deepfake, released 120 days before to 60 days after an election, that showed a candidate “doing or saying something that the candidate did not do or say.”[10] In response, Judge Mendez blocked this ban because it lacked free-speech protections under the First Amendment for parody and satire.[11] Judge Mendez said, “It’s become a censorship law and there is no way that is going to survive.”[12] Accordingly, the First Amendment presents the primary challenge to laws that regulate political ad content.[13]
III. The Tension Between the First Amendment and AI Regulation
“In general, the government cannot regulate the content of candidates’ political ads” under the First Amendment.[14] As a result, candidates and their supporters have attempted to deceive voters through advertising ever since the dawn of elections in this country.[15] In keeping with that tradition, deepfakes are the most recent way to sway the public with disinformation in the Age of AI.[16] For instance, months before the 2024 presidential election, Elon Musk shared a video on X that satirized a real “Harris for President” campaign ad.[17] In the fake ad, an AI-generated version of Harris’s voice claims that she became the Democratic candidate for president because of her race and gender, as well as President Biden’s perceived senility.[18] This deepfake highlights the difficulty of regulating the use of high-quality AI tools in political advertising, particularly when the ad is First Amendment-protected satire.[19]
As it stands, the only regulatory guides are social-media-platform rules[20] and a patchwork of state AI-disclosure laws that require disclaimers on deepfake political ads.[21] As of July 2025, twenty-four states, including California, require the disclosure of deepfake content in political advertising.[22] Regardless, the impact of deepfake video is no more alarming than the impact of misleading headlines or audio.[23] In two surveys of the same representative sample of Americans, deepfake videos, misleading textual headlines, and audio recordings each had a deception rate of over 40%.[24] Therefore, even if AI-disclosure laws proved to be effective, political advertisers could simply deliver their most potent misleading content in headline or audio form to avoid disclosure.[25]
IV. Combating Political Deepfake Ads
There is, however, some recourse for those who have been harmed by political disinformation.[26] “If the deepfake is defamatory, invades privacy, or results in emotional distress, individuals can pursue legal remedies under existing laws covering defamation and privacy violations.”[27] Even still, these legal remedies apply only after the deepfakes have spread online without any warning that the deepfakes contain deceptive content.[28]
As a result, for political candidates without enough resources to fend off deepfake attacks, “[l]ocal-level elections will be so much more challenging because people will be attacking [those candidates via AI tools].”[29] In 2022, when the mayor of Shreveport, Louisiana, ran for reelection, a rival political action committee used AI-generated video “to [falsely] depict [the mayor] as a high school student who had been called into the principal’s office. Instead of giving a tongue-lashing for cheating on a test or getting in a fight, the principal blasted [the mayor] for failing to keep communities safe and create jobs.”[30] Despite the AI disclosure included in the ad, the ad resonated with voters, and the mayor lacked the resources and staff to counteract it.[31] After losing the race, he realized just how significant the ad had become.[32] Disclosure did not save him.[33]
V. Conclusion
As the 2026 midterm elections draw near, it is up to each of us to be skeptical of everything we see and hear[34] because no law currently on the books can fully protect us from the effects of misleading political ads.[35] The focus on regulating AI in politics is, at least in part, an attention-grabbing opportunity for powerful politicians to appear invested in the prevention of political dirty tricks, but no one has ever successfully stopped these deceptive tactics throughout the history of the United States.[36] Tolerating political deception, whether AI-generated or not, is just one of the many sacrifices we have to make to enjoy our constitutional freedom to say almost anything we want.[37]
*Chauncey Bellamy is a second-year student at the University of Baltimore School of Law where he is a Staff Editor for the Law Review, a Teaching Assistant for Introduction to Lawyering Skills, and a Distinguished Scholar of the Royal Graham Shannonhouse III Honor Society. He received a Bachelor of Fine Arts in Theatre with honors from New York University and spent this past summer as a summer associate at Whiteford, Taylor & Preston. Next summer, he will be a summer associate at Ballard Spahr.
[1] See Adam Wren, How Gavin Newsom Trolled His Way to the Top of Social Media, Politico (Aug. 20, 2025, at 10:04 ET), https://www.politico.com/news/2025/08/20/gavin-newsom-twitter-trump-00515785.
[2] Id.; see image posted by Governor Newsom Press Office (@GovPressOfice), X, so nice! (Aug. 17, 2025, at 11:57 ET), https://x.com/GovPressOffice/status/1957109612211229007.
[3] See image posted by Governor Newsom Press Office (@GovPressOffice), X, WOW! WHAT AN HONOR! ON MOUNT RUSHMORE, THANK YOU!! — GCN (Aug. 18, 2025, at 10:34 ET), https://x.com/GovPressOffice/status/1957451025310199852 (reposted from Meacham, @MeachamDr).
[4] See Wren, supra note 1.
[5] Id.
[6] See Richard Sill, Free Speech Rights Secure a Legal Victory over California’s Restrictive Deepfake Laws, Reason (Aug. 21, 2025), https://reason.org/commentary/free-speech-rights-secure-a-legal-victory-over-californias-restrictive-deepfake-laws/; see also Dan Walters, Deepfakes Pose an Obvious Peril in Politics, But California’s Bans Amount to Censorship, CalMatters (Aug. 7, 2025), https://calmatters.org/commentary/2025/08/deepfake-politics-california-law-censorship/; Ali Swenson, Dan Merica & Garance Burke, How Deepfake Tech Is a High Risk, High Reward Tool for Local Political Campaigns, PBS News (June 17, 2024, at 15:23 ET), https://www.pbs.org/newshour/politics/how-deepfake-tech-is-a-high-risk-high-reward-tool-for-local-political-campaigns.
[7] Sill, supra note 6.
[8] Id.
[9] Walters, supra note 6.
[10] Sill, supra note 6 (quoting Cal. Elec. Code § 20513(a)(2)(A) (West 2025)).
[11] Id.
[12] Chase DiFeliciantonio, Elon Musk and X Notch Court Win Against California Deepfake Law, Politico (Aug. 5, 2025, at 20:32 ET), https://www.politico.com/news/2025/08/05/elon-musk-x-court-win-california-deepfake-law-00494936.
[13] See Can Candidates Lie in Political Ads? A First Amendment Analysis, Freedom F., https://www.freedomforum.org/lie-political-ads/ (last visited Sep. 27, 2025).
[14] Id.
[15] Elaine Kamarck, A Short History of Campaign Dirty Tricks Before Twitter and Facebook, Brookings (July 11, 2019), https://www.brookings.edu/articles/a-short-history-of-campaign-dirty-tricks-before-twitter-and-facebook/.
[16] See Sara Savat, Political Deepfake Videos No More Deceptive than Other Fake News, Research Finds, WashU: The Source (Aug. 19, 2024), https://source.washu.edu/2024/08/political-deepfake-videos-no-more-deceptive-than-other-fake-news-research-finds/.
[17] Ali Swenson, A Parody Ad Shared by Elon Musk Clones Kamala Harris’ Voice, Raising Concerns About AI in Politics, AP News (July 29, 2024, at 13:01 ET), https://apnews.com/article/parody-ad-ai-harris-musk-x-misleading-3a5df582f911a808d34f68b766aa3b8e.
[18] Id.
[19] Id.
[20] Id.
[21] Id.; Artificial Intelligence (AI) in Elections and Campaigns, NCSL (July 23, 2025), https://www.ncsl.org/elections-and-campaigns/artificial-intelligence-ai-in-elections-and-campaigns.
[22] NCSL, supra note 21.
[23] Savat, supra note 16.
[24] Id.
[25] See id. (stating that “disinformation will continue to be a challenge for campaigns” even though deepfakes “are not uniquely powerful at deception or affective manipulation”).
[26] See Elizabeth Williamson, From Pizzagate to the 2020 Election: Forcing Liars to Pay or Apologize, N.Y. Times (Apr. 3, 2024), https://www.nytimes.com/2024/03/31/us/politics/defamation-lawsuits-michael-gottlieb.html.
[27] Sill, supra note 6.
[28] See id.
[29] Swenson, Merica & Burke, supra note 6 (quoting Zelly Martin, senior research fellow at the University of Texas at Austin’s Center for Media Engagement and lead author of the university’s research report on the use of AI tools by political consultants).
[30] Id.
[31] Id.
[32] Id.
[33] See id.
[34] Brendan Rascius, The Charlotte Observer, How to Spot Political Deepfake Ads This Year, Gov’t Tech. (Feb. 21, 2024), https://www.govtech.com/artificial-intelligence/how-to-spot-political-deepfake-ads-this-year.
[35] See Swenson, Merica & Burke, supra note 6.
[36] See Kamarck, supra note 15.
[37] See Freedom F., supra note 13.
University of Baltimore Law Review 