29 States Now Regulate AI Deepfakes in Elections

American voters face dramatically different levels of protection against AI-generated deepfakes depending on where they live, as states forge ahead with their own regulations while federal lawmakers remain gridlocked.

Twenty-nine states have implemented laws governing AI deepfakes in elections, according to the National Conference of State Legislatures. However, the approaches vary widely—from outright bans during specific periods to disclosure requirements of varying stringency.

Why it matters

AI-generated political content has become widespread in the current midterm election cycle, with candidates across the political spectrum deploying synthetic media in campaign materials. Without consistent national standards, voters in different states encounter vastly different regulatory environments, potentially affecting their ability to identify manipulated content.

State approaches diverge sharply

The regulatory strategies fall into distinct categories. Minnesota and Texas have adopted time-limited prohibitions, banning political deepfakes for a specified number of days before elections. Maryland takes a more comprehensive approach with a year-round ban.

Most other states with laws on the books require disclosure when AI is used in political advertising. The specificity of these requirements varies considerably—Colorado and Utah mandate detailed information including the creator's identity, creation date, and editing methods.

Court challenges reshape the landscape

Two states have seen their deepfake laws permanently blocked on First Amendment grounds. California and Hawaii have both had their AI election deepfake regulations struck down in court, forcing them to pursue alternative regulatory paths.

California's response includes the AI Transparency Act, which partially took effect on August 3 and requires disclosures for AI-generated images. Hawaii passed legislation last month prohibiting deepfakes in advertisements without the subject's consent—a different legal framework that sidesteps the election-specific restrictions that courts found problematic.

Federal action remains limited

Congress has yet to establish baseline standards for election-related AI deepfakes, despite years of proposed legislation. The only federal deepfake law enacted so far addresses a different concern: the Take it Down Act, which became effective in May, targets non-consensual intimate imagery.

Implementation of that law has proven challenging, according to advocates who cite inconsistent content removal processes and predict litigation will be necessary to achieve robust enforcement. The issue has intersected with election politics through deepfake attacks targeting women lawmakers.

House Democrats have indicated plans to pursue election deepfake legislation if they regain power next year, building on the initial framework established by the Take it Down Act.

The fragmented state-by-state approach reflects a broader pattern: while federal proposals have languished, state legislatures have moved decisively, creating a patchwork of protections that leaves voters in different jurisdictions with fundamentally different safeguards against synthetic media manipulation.

These details were first reported by Axios.