New federal crime: AI deepfakes impersonating government officials

What it does

This bill amends federal law to make it illegal to use artificial intelligence to impersonate a federal official—such as by generating a fake video of a government employee speaking—without a clear disclaimer. The law allows up to 3 years in prison and fines. It explicitly protects satire, parody, and other First Amendment speech if the content clearly discloses it is not authentic.

Why we flagged it

The bill adds a new criminal offense to 18 U.S.C. § 912, targeting a specific form of fraud (AI-based impersonation of federal officials). It is a straightforward public-safety measure with no private beneficiaries or carve-outs.

What the text implies

• The definition of 'artificial intelligence' is broad ('any system or software that performs tasks normally requiring human intelligence') and could capture non-generative tools; enforcement discretion will determine whether simple voice-changing software or deepfakes fall within scope.
• The 'materially false or misleading' standard requires prosecutors to prove both that content was deceptive AND that it was reasonably likely to cause belief in authenticity; this may create litigation over what constitutes 'material' falsity in political or satirical contexts.

The full analysis lists 4 implications of this text.

Who it affects

Citizens gain protection against a novel form of fraud and deception—AI-generated impersonations of federal officials that could undermine trust in government, enable scams, or spread disinformation. The law preserves First Amendment rights by explicitly protecting satire and parody with disclosure, so ordinary speech and political commentary remain protected.