US Enforces Federal Deepfake Law for First Time
A new US federal law criminalizing malicious deepfakes has been enforced for the first time, signaling a major shift in legal risk for AI-generated content. The move comes as high-profile figures like Paris Hilton are spearheading advocacy for stronger protections, and globally, prosecutors in South Korea are seeking a 3-year hard labor sentence for a streamer over deepfake charges.
The criminal provisions of the TAKE IT DOWN Act, signed into law in May 2025, are now in effect. This legislation marks the first federal law in the U.S. to directly criminalize the creation and distribution of non-consensual intimate imagery (NCII), which explicitly includes AI-generated deepfakes. Penalties for offenders can include up to two years in prison for content involving adults and up to three years for content depicting minors. The act creates a dual-accountability framework, targeting both creators and platforms. Beyond individual criminal liability, the law mandates that social media companies and other websites establish a process for victims to request the removal of such content. Platforms have a 48-hour window to comply with a victim's takedown notice, with enforcement of these platform obligations handled by the Federal Trade Commission (FTC). Complementing this criminal statute is the pending DEFIANCE Act, which would grant victims a federal civil remedy. Championed by Paris Hilton alongside a bipartisan group of lawmakers including Rep. Alexandria Ocasio-Cortez, this bill would empower survivors to sue creators and distributors of non-consensual sexual deepfakes for significant financial damages. This federal action builds on a complex patchwork of state-level legislation. While nearly all states have some laws against non-consensual imagery, many were enacted before the rise of AI and do not specifically cover digital forgeries. States like Texas and Virginia have been pioneers, passing laws that specifically target deepfakes used to influence elections or create non-consensual pornography. Globally, the U.S. approach of creating specific criminal and civil penalties contrasts with the European Union's strategy. The EU's AI Act focuses primarily on transparency, mandating that AI-generated content, including deepfakes, must be clearly labeled as artificial. This places the burden on disclosure rather than outright criminalization of creation, though distribution can be policed under other regulations like the Digital Services Act.
