H.R. 7786: AI Fraud Accountability Act

H.R. 7786, the AI Fraud Accountability Act, is aimed squarely at scams that use highly realistic AI-generated audio or video to impersonate real people or even made-up people who look and sound real. The bill defines a "digital impersonation" in unusually specific terms: it covers visual or audio depictions made with software, machine learning, artificial intelligence, or other computer-generated means that would be indistinguishable to a reasonable person from a real person, whether the target is an identifiable individual or an imaginary individual presented as real. It also defines an "identifiable individual" as someone whose face, likeness, voice, or other distinguishing characteristic, including a unique birthmark or other recognizable feature, appears or is heard in the fake content.

The core legal change is simple: if someone uses interstate or foreign communications or commerce to falsely pose as an identifiable or imaginary individual in a digital impersonation with intent to defraud a person of money, paper, document, or another thing of value, that conduct becomes illegal. On the criminal side, the penalty is a fine under Title 18, imprisonment for not more than 3 years, or both. The bill also covers threats: intentionally threatening to commit this kind of fraud for intimidation, coercion, extortion, or to create mental distress brings the same penalty. Courts would also have to order forfeiture of gross proceeds from the violation and personal property used, or intended to be used, to facilitate it, using procedures under Section 413 of the Controlled Substances Act, 21 U.S.C. 853, except subsections (a) and (d). The bill also applies extraterritorial federal jurisdiction, meaning cases can reach conduct outside the United States.

The measure does not rely only on criminal prosecution. It also gives the Federal Trade Commission authority to enforce a parallel civil ban, treating violations as unfair or deceptive acts or practices under Section 18(a)(1)(B) of the Federal Trade Commission Act. That matters because many scams move fast and hit lots of people at once, so regulators may need tools beyond criminal cases. At the same time, the bill includes carveouts for lawfully authorized investigative, protective, or intelligence activities by federal, state, or local law enforcement and U.S. intelligence agencies, plus a savings clause making clear that parody, satire, journalism, and other First Amendment-protected rights are not restricted.

The back half of the bill is about building a long-term response. Not later than 30 days after enactment, NIST would have to convene a working group with DOJ, FTC, law enforcement, private-sector industries like finance, health care, retail or e-commerce, telecommunications, and digital platforms, along with scientists and engineers in digital forensics and AI. That group must publish best practices not later than 1 year after enactment, update them not later than 2 years after enactment and at least annually after that, and send Congress a report not later than 1 year after enactment and annually thereafter. It sunsets 10 years after enactment. Separately, within 90 days after enactment, the FTC must identify the top 10 foreign countries where these violations originate and harm U.S. persons, then report to Congress within 1 year and annually after that, while DOJ reviews international agreements within 1 year and every 5 years thereafter.