The 2026 revision of the AI Act

IRIS 2026-7:1/11

Justine Radel-Cormann

European Audiovisual Observatory

On 29 June 2026, the Council of the European Union approved the regulation amending the existing AI Act. The text was officially signed on 8 July 2026 by both the President of the European Parliament and the President of the Council, and is soon to be published in the Official Journal of the European Union. 

With this revised version, stronger protections for children in the context of AI have been implemented. Legislators introduced new prohibitions in Article 5 targeting certain AI practices. These provisions are set to take effect from 2 December 2026. It is now forbidden to place on the market, put into service, or use an AI system that generates or manipulates realistic images, videos, audio or similar material of an identifiable natural person’s intimate parts, or of an identifiable natural person engaged in sexually explicit activities, without that person’s freely given, specific, informed, unambiguous and explicit consent for such generation or manipulation. 

Additionally, the amended Act prohibits AI systems that generate or manipulate materials or performances involving child pornography, or pornographic performances by a child, except in cases where a "without right" defence applies under national law. In the context of child pornography, the term "without right" allows member states to provide a defence in respect of conduct relating to pornographic material having, for example, a medical, scientific or similar purpose.

The modifications clarify that these prohibitions apply only when the generation or manipulation of such materials or performance is the intended purpose of the AI system, or when the system’s design, training, architecture, capabilities or user-facing functionalities make that generation or manipulation a reasonably foreseeable and reproducible outcome, without requiring significant technical modification. These prohibitions also apply when the system lacks reasonable and adequate safety measures and other safeguards to reliably prevent such generation or manipulation, taking into account reasonably foreseeable misuse, and to correct observed or reported misuse. The prohibitions also only apply when a deployer uses the system for the purpose of generating or manipulating such materials or performances. 

Moreover, transparency requirements under Article 50 have been strengthened. Providers of AI systems, including general-purpose AI, must ensure that synthetic audio, image, video, or text content generated by their systems is marked in a machine-readable format and is detectable as artificially generated or manipulated. The revised Article 113 stipulates that AI systems already on the market before 2 August 2026 must take the necessary steps to comply with the updated Article 50 requirements by 2 December 2026. 

The text includes other changes aimed at simplifying the AI Act, such as the clarification of the competences of the AI Office, the delayed establishment of AI regulatory sandboxes, or the delayed application of the high-risk AI systems rules.