Federal Deepfake Takedown Law Takes Effect This Month

The TAKE IT DOWN Act's compliance deadline arrives on May 19. From that date, online platforms operating in the United States are required to remove non-consensual intimate images, including AI-generated deepfakes, within 48 hours of receiving a valid notice from the depicted person.

The law was signed by President Donald Trump on May 19, 2025, and gave covered platforms one year to build the notice-and-takedown infrastructure. The criminal provisions, which make knowingly publishing such material a federal offence, took effect at signing. The platform obligations are what activate this month.

What the law actually requires

A "covered platform" is broadly defined and includes most consumer-facing websites and apps that host user-generated content. To submit a takedown request, the depicted individual must provide an electronic signature and a brief statement saying, in good faith, that the image was published without their consent. The platform then has 48 hours to investigate and remove the material from its services, and must make reasonable efforts to identify and remove copies.

The Federal Trade Commission has enforcement authority. A failure to comply is treated as a violation of the FTC Act, which carries civil penalties and injunctive relief.

The AI dimension

What sets the TAKE IT DOWN Act apart from earlier non-consensual imagery laws is its inclusion of "digital forgeries." The statute defines this as an intimate visual depiction of an identifiable individual created or altered using AI or other technological means. According to a Nelson Mullins analysis, this is the first federal law that limits the use of AI in ways that can be harmful to individuals.

The deepfake clause arrives as image generation has become trivially accessible. Free and low-cost open-source models can produce convincing intimate imagery of real people from a handful of public photos, a problem schools and universities have flagged repeatedly over the past two years.

What's coming next

Several state laws on synthetic media follow the federal Act. The Colorado AI Act takes effect on June 30, 2026. California's CCPA provisions on automated decision-making come into force in January 2027. The Trump administration's National Policy Framework for Artificial Intelligence, released in March, recommends that Congress preempt state-level AI laws to avoid a patchwork. That fight is still in Congress.

For platforms, however, the immediate task is operational: build a takedown form, train a review team, and start the clock when notices come in.

Sources: Skadden, Latham & Watkins, Wikipedia.