How are AI deepfakes and AI-generated content regulated?

There is no single legal definition. The European Union's AI Act defines a deep fake as AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and "would falsely appear to a person to be authentic or truthful."¹ U.S. state laws use terms such as "deepfake," "synthetic media" and "deceptive media," and the National Conference of State Legislatures notes that differences in their definitions affect how the laws work in practice.²

Most rules on election deepfakes come from the states. As of October 2026, the National Conference of State Legislatures counted 31 states with laws on deepfakes in political messaging. Most require a disclosure on the media. California, Minnesota and Texas enacted prohibitions on publishing political deepfakes within a set period before an election, and Maryland prohibits deceptive election deepfakes year-round.²

Several of these laws have been challenged on free-speech grounds. In August 2025 a federal judge ruled that California's law violated the First Amendment and permanently barred the state from enforcing it against the satirists who sued; the legislature later amended the law.³² A federal court in Hawaii found that state's law unconstitutional in January 2026.² In September 2026, federal judges in Vermont and Montana granted narrower preliminary injunctions: Vermont's covers one satirical video,⁴ and Montana's applies only to the political action committee and its leader who sued.⁵ In the California case, the court objected that the law reached speech merely "reasonably likely" to harm a candidate's electoral prospects, and found its separate disclaimer requirement for satire and parody overly burdensome.²

Wisconsin requires political synthetic media meant to influence an election to say that it was generated by AI.² In a 2026 congressional race, after Rep. Derrick Van Orden posted AI videos of his opponent, Rebecca Cooke, Cooke's campaign argued in a cease-and-desist letter that defamation law "fills the gaps yet remaining in Wisconsin's statutory protections against fake AI."⁶

The Federal Election Commission declined in September 2024 to write rules specific to AI in campaign ads, saying it would apply its existing ban on fraudulently misrepresenting campaign authority case by case.⁷ In an interpretive rule the same day, it said that ban applies "irrespective of the technology used," including AI.⁸

AI voices in robocalls fall under the Telephone Consumer Protection Act. In February 2024 the Federal Communications Commission ruled that the law's limits on calls using an "artificial or prerecorded voice" cover AI-generated voices, so such calls need the called party's prior express consent unless an emergency purpose or exemption applies.⁹ The month before, thousands of New Hampshire voters had received calls in an AI-generated imitation of President Joe Biden's voice ahead of the state's primary; the FCC later finalized a $6 million fine against the consultant who organized them.¹⁰ In August 2026, the Club for Growth asked the FCC to let political callers use artificial or prerecorded voices, including AI-generated voices, on cellphones without prior consent, subject to call limits and opt-outs.¹¹

The TAKE IT DOWN Act, which became law in May 2025, makes it a federal crime to knowingly publish nonconsensual intimate images of an identifiable person, including AI-generated "digital forgeries" (for images of adults, the publication must be intended to cause harm or cause harm), and requires covered platforms to remove them within 48 hours of a valid request.¹²

Outside elections, many rules come from platforms and other governments. The EU's AI Act requires deployers of AI systems that generate or manipulate deepfakes, meaning users other than people acting in a personal, non-professional capacity, to disclose that the content is artificial, starting August 2, 2026, with a lighter requirement for evidently artistic, creative, satirical or fictional works.¹

Valve requires developers selling games on Steam to describe AI-generated content in a content survey and shows much of each disclosure on the game's store page.¹³ Its current guidance says the disclosure concerns AI-generated content that ships with a game and is consumed by players, such as artwork, sound, narrative and localization, rather than efficiency gains from AI development tools.¹⁴ Bandcamp bars music and audio "generated wholly or in substantial part by AI" and prohibits using AI to impersonate other artists.¹⁵

Authorship is a separate question. The U.S. Copyright Office concluded in January 2025 that copyright protects a human author's original expression even when a work includes AI-generated material, but does not extend to purely AI-generated material, and that, given current generally available technology, prompts alone do not give users enough control to count as authors.¹⁶

Opponents of loosening the rules include FCC Commissioner Anna Gomez, who said in October 2026 that granting the Club for Growth request would bring "chaos and deception so close to Election Day," pointing to the New Hampshire calls.¹⁰ The Club for Growth argues that "Current TCPA rules chill important political speech."¹¹ In court, challengers to state election deepfake laws have argued that they violate the First and Fourteenth Amendments.²