AI Restrictions in Political Ads: What to Know About “Deepfake” Disclaimers and Bans
The capacity of artificial intelligence (AI) to generate realistic videos and images of people (commonly known as “deepfakes”) is advancing at a rapid pace. In recent elections, political campaigns, super PACs, and political professionals have begun using AI to generate imitations that are almost indistinguishable from the voices and likenesses of real people. Organizations seeking to use these tools in political messaging should be careful to ensure compliance with a litany of new laws in this space.
While federal law does not yet regulate these kinds of ads, restrictions are increasing at the state level. As of June 2026, more than 30 states have passed some type of law regulating AI-generated media that applies to political advertising. These laws generally fall into three categories:
- Disclaimers: This is the most common category of deepfake law, requiring that political advertisements with AI deepfakes contain clear, discernible audio or visual labels that indicate the content is AI-generated. New Mexico and Mississippi, for example, have deepfake disclaimer laws. Some states, such as Colorado and Washington, also require labels on the content’s metadata (“metadata tags”) in addition to the content a viewer hears and sees onscreen. Aside from state-level restrictions, most social media platforms (e.g., Facebook, Instagram, YouTube) also have disclaimer or disclosure rules that apply specifically to political ads.
- Blanket Prohibitions: Some states have acted more broadly and generally prohibit the use of any artificial or synthetic media or audio that realistically impersonates an individual with the intent to deceive voters. For example, Maryland and Texas all have such laws in place.
- Quasi-Property Protection/Right of Publicity: Other states have established quasi-property rights for individuals to their visual likeness and voice. This provides a right of action for individuals to sue anyone who uses AI to depict them without their permission. For example, in Montana and Virginia, a candidate depicted in an AI-generated imitation could sue as they would for a trademark or copyright infringement. While these laws are typically of general applicability, some states exempt political campaigns.
Enforcement of these laws varies by state. In most states, candidates who are depicted by AI deepfakes can sue whoever paid for or sponsored that ad to get it taken down. Some states, like California, allow anyone (not just the candidate) to sue to get an AI deepfake taken down. In a few states, such as Maryland, only the state can enforce AI deepfake laws and there is no private right of action. Other states have penalties for violators, which range from small fines to felonies and jail time.
States provide various exemptions to their AI deepfake laws. Generally, the creation of AI deepfakes is permitted for use in satire and parody. AI deepfake laws have recently been overturned in California and Hawaii over failure to include exemptions for satire, in violation of the First Amendment. Additionally, most states permit news broadcasters to include AI deepfakes in bona fide news reporting, though those states often require those broadcasters to include a disclaimer when the authenticity of a video or audio is suspected to be AI-generated.
Ultimately, makers of political ads will need to stay vigilant and closely monitor which laws apply to jurisdictions in which they seek to advertise. Wiley’s Election Law & Government Ethics team can help you navigate this rapidly evolving area of law.
Matt Heller, a Summer Associate at Wiley Rein LLP, contributed to this alert.
