Sexual deepfakes, meaning AI-made intimate images of real people shared without consent, are now a specific crime in both the United States and Canada. The US TAKE IT DOWN Act was signed on May 19, 2025 and made platform removal duties enforceable on May 19, 2026, while Canada's Bill C-16 received Royal Assent on June 18, 2026 and its Criminal Code section 162.1 now names AI software.
Key takeaways
- The US law makes it a federal crime to knowingly publish an intimate "digital forgery" of an identifiable person without consent, with up to two years in prison for adult victims and up to three for minors.
- Covered US platforms must remove a reported image within 48 hours of a valid request, and the Federal Trade Commission (FTC) enforces that duty.
- Canada's section 162.1 covers AI-made images that are likely to be mistaken for real recordings, with up to 10 years on indictment.
- Both regimes target publishing and threats; neither is a general ban on image-generation tools.
Why this harm gets its own laws
Many risks attributed to AI are broad or contested. This one is concrete: a model can produce a realistic intimate image of a named person from ordinary photos, and the image can spread faster than a victim can find it. The National Center for Missing and Exploited Children states that its CyberTipline received more than 400,000 reports with a generative AI connection in 2025, out of 21.3 million reports in total, and that more than 182,000 of the 400,000 involved offenders possessing, generating, or trying to generate AI-made child sexual abuse material. These are minimum counts of reports, which is not the same as counts of victims.
United States: crimes and a removal duty
The TAKE IT DOWN Act (Public Law 119-12) has two parts. Section 2 adds a criminal offense to 47 U.S.C. 223. For adults, it is unlawful to knowingly publish an intimate "digital forgery" through an interactive computer service without the person's consent, where publication is meant to cause harm or does cause psychological, financial, or reputational harm. A digital forgery is an intimate depiction of an identifiable person made with software, machine learning, AI, or other technological means, which a reasonable viewer would find indistinguishable from an authentic depiction.
Penalties are a fine, up to two years in prison, or both for adult victims, and up to three years for minors. Threatening to publish a forgery carries up to 18 months for adults and 30 months for minors. Courts must order forfeiture and restitution.
Section 3 applies to covered platforms: public sites, services, or apps that primarily host user-generated content. After a written request with a signature, location details, a good-faith statement of nonconsent, and contact information, the platform must remove the image within 48 hours and make reasonable efforts to remove known identical copies. Email and broadband access services are excluded. The FTC treats failure to comply as an unfair or deceptive practice.
FTC enforcement since May 2026
The FTC announced that it began enforcing section 3 on May 19, 2026, the date platforms were required to have a request process in place. Its chairman sent reminder letters the week before to 15 major platforms, including Alphabet, Meta, Microsoft, and X. People can file complaints about platforms that ignore valid requests or lack a process, through the FTC's Take It Down complaint page. The FTC announcement does not mention penalties, and this review found no enforcement action against a named platform.
Canada: section 162.1 after Bill C-16
Bill C-16, the Protecting Victims Act, received Royal Assent on June 18, 2026 (Statutes of Canada 2026, chapter 19). The Justice Laws version of Criminal Code section 162.1 read on October 24, 2026 defines an "intimate image" to include a visual representation made by electronic or mechanical means, "including by means of artificial intelligence software," that shows an identifiable person nude, nearly nude, or in explicit sexual activity and is likely to be mistaken for a real recording. Sharing such an image without consent, or being reckless about consent, is punishable by up to 10 years on indictment, and the offense can also be prosecuted summarily. Threatening to share it with intent to intimidate, or to be taken seriously, carries the same maximum. The section has a public-good defense. Law firm commentary reports that the earlier wording reached only genuine recordings and that most of the bill's Criminal Code changes took effect July 18, 2026.
Side by side
| Feature | United States | Canada |
|---|---|---|
| Instrument | TAKE IT DOWN Act, Public Law 119-12 | Criminal Code s. 162.1, amended through Bill C-16 |
| Enacted | May 19, 2025 | Royal Assent June 18, 2026 |
| AI-made images | "Digital forgery" indistinguishable from an authentic depiction | Image made by electronic means, including AI software, likely to be mistaken for a real recording |
| Maximum for sharing | 2 years (adult victim), 3 years (minor) | 10 years on indictment |
| Platform duty | Remove within 48 hours of valid request | None in the section reviewed |
| Enforcer | Prosecutors; the FTC for platform duties | Prosecutors |
What the texts do not cover
- Neither law was found to bar the making of a tool or the private creation of an image; the offenses concern publishing, distributing, or threatening.
- Both use a realism threshold, so an image no reasonable viewer would take for real may fall outside.
- The US removal duty is triggered by a request, not by platform detection.
- This article covers one documented harm. It does not survey other documented harms, such as discrimination in automated decisions or fraud.
This article is general information, not legal advice. Statutes get amended and are applied case by case, so anyone dealing with a specific image should speak with a lawyer or victim-support service and read the primary texts, including the TAKE IT DOWN Act.
FAQ
Is it illegal to make an AI deepfake of someone?
The provisions reviewed here punish publishing, sharing, or threatening to share. Whether making an image is a separate offense depends on other laws, such as those on child sexual abuse material, and on local rules.
How fast must a US platform remove a reported image?
Within 48 hours after a valid written request, and the platform must also make reasonable efforts to remove known identical copies.
Does Canadian law now cover AI-generated images?
Yes. Section 162.1 expressly mentions artificial intelligence software and covers images likely to be mistaken for real recordings.
Where can a US complaint about a platform be filed?
The FTC points to its Take It Down complaint page for complaints about platforms that ignore valid removal requests or have no request process.
Related reading
- other documented AI harms
- the argument that AI threatens society
- AI harms in education
- AI's wider effect on society
More in AI Policy and Ethics.
Last verified: October 2026
Sources
- TAKE IT DOWN Act, Public Law 119-12, United States Government Publishing Office, May 19, 2025
- FTC Begins Enforcing TAKE IT DOWN Act, Federal Trade Commission, May 2026
- Criminal Code, section 162.1, Justice Laws Website, Government of Canada
- Bill C-16, Protecting Victims Act, LEGISinfo, Parliament of Canada, 2026
- CyberTipline data, National Center for Missing and Exploited Children, 2025 figures
- Sexual Deepfakes and Intimate Images: What Bill C-16 Newly Criminalizes, Karapancev Law, 2026 (commentary on earlier wording and in-force date)