Across the United States, voters may encounter very different levels of protection from AI-generated deepfakes in the run-up to Election Day because state-level rules are inconsistent.
Why this matters
AI-generated attack ads and other campaign content have become widespread ahead of the midterm elections, with candidates across the political spectrum using the technology.
Current state of play
According to the National Conference of State Legislatures (NCSL), 29 states have AI election deepfake laws currently in effect. Two states — California and Hawaii — have had their election-focused laws permanently enjoined by courts.
Some states have adopted stricter limitations: Minnesota and Texas prohibit political deepfakes for a specified number of days before an election, while Maryland's ban applies year-round. Other states focus on disclosure requirements when AI is used in political advertising. A few, including Colorado and Utah, require more detailed disclosures such as the creator of the deepfake, the date it was created, and how it was edited.
Legal friction and alternative approaches
Courts struck down California's and Hawaii's AI deepfake laws on First Amendment grounds. Both states are pursuing alternative measures that do not specifically regulate election deepfakes. California's AI Transparency Act, portions of which went into effect on August 3, 2024, requires labeling of AI-generated images. Hawaii recently passed a law banning deepfakes in ads without the subject's consent.
Federal actions and next steps
At the federal level, lawmakers have not yet established a baseline standard for election AI deepfakes. Congress' first AI-related deepfakes law was the Take It Down Act, which took effect in May 2024 and targets non-consensual intimate imagery — an issue that has intersected with election ads aimed at women lawmakers. Advocates warn that passing the law is only the first step and that inconsistent content-removal processes mean litigation may be needed to secure effective enforcement.
House Democrats have indicated they plan to push legislation next year specifically targeting election deepfakes if they regain power.
Bottom line
Although Congress has discussed proposals to address AI in elections for years, states have moved faster, resulting in a patchwork of protections that leave voters’ safeguards dependent on where they live.
