Michigan's mandatory disclosure law for AI-generated campaign ads is getting its most direct test yet, with candidates in two of the state's marquee 2026 races weaving synthetic media into their pitches to voters. Republican gubernatorial candidate Aric Nesbitt, a former state senator, released a campaign video in June built around AI-generated depictions of Governor Gretchen Whitmer and Secretary of State Jocelyn Benson, closing the spot with a synthetic version of Attorney General Dana Nessel reciting the state's own required AI disclaimer. Around the same time, U.S. Senate candidate Mike Rogers went viral for an AI image that gave him a noticeably more muscular physique.
The Nesbitt video, using likenesses of Whitmer and others per the Free Press, folds Michigan's disclosure mandate into the ad itself rather than skirting it — a move that shows how quickly campaigns are adapting to rules state lawmakers wrote back in 2023. As Axios reports, Michigan is now among roughly 29 states with AI election deepfake laws in effect, part of a patchwork that leaves voters in different states experiencing very different rules governing what they see on their screens. Whitmer signed the original four-bill package on November 30, 2023, requiring clear disclosures on political ads substantially created with artificial intelligence, banning deceptive pre-election deepfakes within 90 days of an election, and setting civil fines of up to $250 for a first offense and $1,000 for repeat violations, according to the State of Michigan.
A Law Built for a Slower-Moving Threat
Michigan was among the first five states in the country to require disclaimers on AI-generated political ads ahead of the 2024 election cycle, a distinction noted by Broadcast Law Blog. But the lead sponsors of that 2023 law have acknowledged that rapidly advancing generative tools present ongoing enforcement challenges, per reporting from Bridge Michigan. Michigan's law requires disclosure whenever AI or materially deceptive media is used in campaign ads, and many other states now have similar disclosure requirements on the books.
Maryland has gone further, prohibiting political deepfakes year-round, while Minnesota and Texas restrict them for a set number of days before an election. The Michigan attorney general has said that AI use makes it harder for voters to determine fact from fiction, according to the Axios report — a concern that sits at the center of why states have rushed to legislate even as Congress has not established a baseline federal standard for election-related AI deepfakes.
Courts Are Already Pushing Back
That legislative rush is colliding with the courts. Of the roughly 29 state election-AI deepfake laws in effect as of June 23, 2026, two similar laws have already been struck down by courts, according to the Axios count. The clearest example came in California, where Senior U.S. District Judge John A. Mendez issued a preliminary injunction in October 2024 and then struck down the state's pre-election deepfake law, AB 2839, in September 2025, ruling that pre-election bans on deceptive political deepfakes unconstitutionally restrict speech based on content and viewpoint, as reported by Tech Policy. That case followed a lawsuit from a content creator who had published AI parody videos of presidential candidates.
The ruling underscores a constitutional vulnerability facing state AI restrictions nationwide, and it helps explain why the AI election landscape now varies so widely depending on where a voter lives. Legislative tracking shows that 31 states had enacted laws regulating political deepfakes or requiring campaign AI disclaimers by mid-2026, up from just five states in late 2023 — a rapid expansion that has unfolded largely without federal guidance.
Washington Has Largely Stayed on the Sidelines
At the federal level, the Federal Election Commission voted 5-1 in September 2024 against launching a formal rulemaking process for AI campaign ads, opting instead for an interpretive rule affirming that existing statutory bans on fraudulent misrepresentation apply to AI content on a case-by-case basis, per the FEC. That decision came after Public Citizen had formally petitioned the agency in 2023 to establish explicit AI election regulations. Separately, the Federal Communications Commission issued a Notice of Proposed Rulemaking in July 2024 seeking to mandate on-air and written disclosures whenever AI-generated content appears in television and radio political ads, though its jurisdiction stops short of covering online platforms or streaming services.
Congressional proposals to tackle AI in elections have been floated for years without resulting in a comprehensive federal law. The one major exception passed with bipartisan support: the TAKE IT DOWN Act, which President Trump signed on May 19, 2025, establishing federal criminal penalties for sharing nonconsensual intimate imagery — including AI digital forgeries — and creating platform notice-and-takedown obligations that took effect in May 2026, according to Latham & Watkins. It marked Congress's first enacted federal deepfake law, and per the Axios report, its focus on nonconsensual intimate imagery has increasingly collided with election ads targeting women lawmakers.
What This Means Heading Into November
Senator Brian Schatz reintroduced the bipartisan AI Labeling Act in July, proposing mandatory human-readable disclosures and embedded metadata provenance tags for AI-generated media across digital platforms — an effort Hoodline previously covered. Supporters argue that metadata provenance is necessary to trace and verify synthetic media online, but the bill still leaves states to fill the regulatory gap in the meantime.
With states having moved faster than Congress on election AI regulation, Americans could experience meaningfully different AI election environments ahead of the November 3 election, depending on which state's rules — and which court rulings — apply where they live. Michigan's own experience this summer, with campaigns folding required disclaimers directly into AI-generated ads, suggests candidates are adapting to disclosure laws rather than avoiding them, even as AI-generated attack ads and campaign content remain rampant heading into the midterms.
