On 19 May 2025 the United States signed the TAKE IT DOWN Act. It addresses non-consensual distribution of intimate images, expressly including synthetic content generated by artificial intelligence. It converts a problem once handled mainly through platform self-regulation into a statutory duty with a defined deadline.
Forty-eight hours: an enforceable deadline
The core mechanism is notice and takedown: on receiving valid notice from a victim or a representative, a platform must remove the material within 48 hours. The deadline matters because it turns the vague phrase “as soon as possible” into a verifiable metric.
For platforms, that means building a verifiable process for intake, identification and removal, rather than relying on the chance responsiveness of manual review.
Why AI-generated content is caught
Traditional governance of intimate images depended on identifying the original material. Synthetic content changes the premise: the image may never have existed, yet the victim suffers the same harm. See our note on the global statement on privacy in AI-generated images.
By placing synthetic content and real images in the same regulatory framework, the Act signals that how content was generated does not affect the finding of harm. That aligns with recent deepfake legislation in a number of countries. See our note on Korea’s punitive damages for AI fake advertising.
How it connects to labelling rules
A takedown duty handles what happens afterwards; a labelling regime handles identification beforehand. The two need to work together: if synthetic content already carries a label at generation and distribution, platforms can identify and remove it far more efficiently. See our notes on the Measures for Labelling AI-Generated Synthetic Content and on users’ verification duty for AI-generated content.
Where overseas-facing companies can start
First, build a dedicated intake channel for intimate images and synthetic content, with an explicit response deadline.
Second, add synthetic-trace detection to content review, shortening the time from notice to takedown.
Third, watch the differences in deadlines between jurisdictions, consolidating multi-jurisdiction takedown duties into a single handling standard. See our note on the US AI Action Plan.
The Act does not require platforms to police content proactively, and it does not set a standard of perfect accuracy. What it fixes is the response time once a valid notice has arrived — a narrow but enforceable obligation that leaves platforms with no room to argue about whether the material should come down.
The hard part of deepfake governance is not legislative statement but whether handling can be completed within an hour-scale deadline. Those who turn process into capability are the ones who comply.
This article is based on the US TAKE IT DOWN Act and public information and is for reference only; it does not constitute legal advice.