What does the law say when your own face starts saying things you never said? It has several answers. They don't all agree with each other.
Deepfake impersonation law is really several laws wearing one name. Impersonation for fraud is illegal everywhere, deepfake or not. Someone fakes you to steal money or access, that's fraud, full stop. The AI part just makes the evidence weirder. Commercial use of your likeness without consent is actionable in many states under right-of-publicity laws, written for lookalikes and soundalikes and now covering synthetic ones.
The gray zone is non-commercial impersonation. Parody accounts, satire, political commentary. Some states protect it robustly. Others have narrower shields, and I don't trust anyone who draws that line with total confidence. Voice clones deserve their own mention. A synthetic voice saying things the person never said sits at the intersection of fraud, likeness, and wiretapping-era laws, and regulators are paying attention. AI-generated likeness law penalties are also growing as their own category, with newer statutes setting specific fines and jail time for realistic fake depictions of real people without consent.
Practically, the question is always who you are in the scenario. Victims have more avenues than they think, spread across fraud, likeness, harassment, and newer deepfake-specific statutes. Creators have fewer safe harbors than they assume, especially once money or elections enter the picture. When in doubt, get consent in writing. Cheapest insurance in this entire area of law. A deepfake law compliance checker tool helps product teams check likeness rules before a feature ships.
DeepfakeLaw maps impersonation and likeness statutes across 91 tracked laws, with penalties and effective dates. And if your product lets users generate likenesses of real people, get a plain-language compliance audit before you ship, at deepfakelaw.fyi.