From Wednesday 2 December 2026, the EU AI Act bans AI systems that create realistic sexual images, video or audio of identifiable people without their explicit consent, the so-called nudify apps, and AI systems that generate child sexual abuse material. On the same date, generative AI services already on the EU market before 2 August 2026 lose their grace period and must mark their output as AI-generated in a machine-readable way. Breaching the ban can cost up to EUR 35 million or 7% of worldwide turnover.

Both changes come from the Digital Omnibus on AI, Regulation (EU) 2026/1744, in force since 27 July 2026 and directly applicable in all 27 member states. The rules below are stated as of October 2026.

What changes on 2 December 2026

  • Nudify ban (Article 5(1)(ba)): no AI system may be placed on the market, put into service or used to generate or manipulate realistic images, video or audio of an identifiable person's intimate parts or sexual activity without that person's freely given, specific, informed, unambiguous and explicit consent.
  • Child abuse material ban (Article 5(1)(bb)): the same applies to AI systems that generate or manipulate child sexual abuse material, unless a "without right" defence applies under national law, for example in criminal investigations.
  • Marking deadline: providers of generative AI systems placed on the market before 2 August 2026 must comply with the marking duty in Article 50(2) from 2 December 2026, under a four-month transitional period added by the Omnibus.

The Commission counts the new bans as a ninth prohibited practice; see our list of AI Act prohibited practices.

Who the nudify ban covers

Under the new Article 5(1a), a provider, including one outside the EU serving the EU market, breaches the ban if generating such material is the system's intended purpose, or if it is a reasonably foreseeable and reproducible result without significant technical modification and the system lacks reasonable and adequate safeguards to prevent it and correct misuse. The second test can reach general-purpose generators; the recitals cite refusal training, prompt guardrails and abuse detection as safeguards.

Users are caught as deployers when they use a system to create such material, but the AI Act's deployer definition excludes personal, non-professional use.

Outside the ban: material made with explicit consent, cartoonish or physically impossible depictions, unidentifiable people, and edits that do not increase exposure, such as changing the background or brightness.

What the marking deadline means

Article 50(2) requires providers of generative AI systems to mark output in a machine-readable format that makes it detectable as AI-generated or manipulated; assistive functions for standard editing are exempt. Systems launched since 2 August 2026 are already covered, and the Commission's Q&A confirms that older ones must comply from 2 December 2026. The grace period covers marking only: deployers have had to disclose deepfakes, and AI-generated text published on matters of public interest, since 2 August 2026.

The voluntary Code of Practice on Transparency of AI-generated Content, published on 10 June 2026 and assessed as adequate by the Commission and the AI Board, shows how to comply. Section 1 covers marking and detection of AI-generated audio, image, video and text: machine-readable marking that is effective, interoperable, robust and reliable as far as technically feasible. Section 2 covers deepfake labelling, where deployers may use the EU icon, which is optional. About 190 organisations, including Google, Meta, Microsoft, Mistral and OpenAI, had signed by the end of July 2026, according to the Commission.

Fines and enforcement

Breaching an Article 5 ban: up to EUR 35 million or 7% of worldwide annual turnover, whichever is higher, from day one, as the penalty rules have applied since 2 August 2025. Breaching Article 50: up to EUR 15 million or 3%. For SMEs and start-ups, the AI Act applies the lower of the two figures.

National market surveillance authorities enforce the rules. The European AI Office supervises AI systems built on a general-purpose model from the same provider or group, and those integrated into very large online platforms and search engines.

How to report

  • A nudify app or non-compliant AI tool: under Article 85 of the AI Act, anyone can complain to a market surveillance authority; the Commission lists the national contact points.
  • An image or video posted online: use the platform's reporting tool, required by the Digital Services Act, or your national Digital Services Coordinator if the platform fails to act.
  • Child sexual abuse material: report it to the police or to the national hotline in the INHOPE network, sending only the link, never the file.

Victims can also go to the police, as such conduct may be a crime under national law. Other dates are in our EU AI Act timeline; our AI Act and policy sections follow enforcement.

Frequently asked questions

Are nudify apps banned in the EU?

Yes, from 2 December 2026. The AI Act prohibits offering, or professionally using, AI systems that create realistic sexual images, video or audio of an identifiable person without explicit consent.

Does the AI Act require watermarks on AI-generated content?

It requires machine-readable, detectable marking without prescribing one technique; the Commission-assessed code of practice sets out machine-readable marking and detection measures.

Do I have to label AI images I post?

Not under the AI Act if you post in a personal, non-professional capacity. Companies, media and other professional users must disclose deepfakes they create with AI.