H.R. 7786: AI Fraud Accountability Act

H.R. 7786, the AI Fraud Accountability Act, targets a specific weapon: AI-generated audio or video realistic enough that a reasonable person can't tell it's fake. The bill calls this a "digital impersonation," and it covers both clones of real people and invented people made to seem real. The trigger is recognizability — a face, a likeness, a voice, even a unique birthmark.

The core rule is short. Use one of these fakes in interstate or foreign communications to pose as someone, with intent to defraud a person of money, papers, documents, or anything of value, and you've broken the law. The penalty is a federal fine, up to 3 years in prison, or both. Threatening to pull off this kind of scam to intimidate, coerce, extort, or cause mental distress carries the same punishment.

A conviction also forces forfeiture: courts must order anyone convicted to hand over the proceeds and any property used to run the scheme. And because so many of these operations sit overseas, the bill claims extraterritorial jurisdiction, letting cases reach conduct outside the United States.

Criminal charges aren't the only lever. The bill gives the Federal Trade Commission authority to enforce a parallel civil ban, treating violations as unfair or deceptive practices under the FTC Act. The bill also carves out lawful law-enforcement and intelligence work, and adds a savings clause stating that parody, satire, journalism, and other First Amendment-protected speech are not restricted.

The rest of the bill builds a long-term response. Within 30 days of enactment, NIST would convene a working group with the DOJ, FTC, law enforcement, and companies in finance, health care, retail, telecom, and digital platforms to develop detection best practices, due within 1 year and updated yearly. Within 90 days, the FTC must name the top 10 foreign countries where these scams originate. The whole working-group effort sunsets 10 years after enactment.