Nobody asks 'is deepfake illegal' and likes the answer. The answer is a question back: where are you, what does the thing show, and who made it. Annoying. Also true.
Rough map, from what I've watched these laws do. Nonconsensual intimate deepfakes are banned in a lot of states now, and the count keeps climbing. Political deepfakes get their own election-season rules in a bunch of places, usually about disclaimers and date windows. Fraud and impersonation mostly ride on old laws that never heard of AI but still hit like a truck.
The part that actually burns people is the spread. Same act, misdemeanor and a fine in one state, felony and jail time in the next. Damage caps differ. Effective dates matter more than people expect, because a law signed in 2024 might only have switched on this year. A few states let you get hit twice, civil lawsuit plus criminal prosecution, which surprises everyone who assumed it was one or the other.
So the question was never the flat one. It's which deepfake laws by state actually touch you, right now. A deepfake legislation database search tool by state beats reading fifty statutes by hand, and DeepfakeLaw tracks 91 of them across all 50 states, federal law, and 40+ countries, with requirements, penalties, and verified in-force dates. If you want the same data pushed to your phone, there's a deepfake law tracker app for iOS and Android on top of it, and an AI disclosure law tracker feed for the labeling rules specifically. Check instead of guessing.
And if you're building with AI-generated media and want someone to map which of these laws actually touch your product, I do plain-language compliance audits at deepfakelaw.fyi.