Member States and the European Parliament reached an agreement last night to ban in the EU artificial intelligence services that allow people to be “undressed” without their consent.

On Tuesday, Italian Prime Minister Giorgia Meloni denounced fake images of herself created using artificial intelligence, calling such deepfakes “a dangerous tool.”

According to the European Parliament, the new EU ban will apply to systems capable of generating pornographic images, videos, or audio involving children, or depicting intimate parts of an identifiable person, or showing them engaged in sexual activities without consent.

The new rules will apply from 2 December 2026. From that date, AI systems will be required to include safeguards to prevent the generation of such content.

The measure was adopted as part of a revision of the European AI Act, a landmark piece of legislation formally adopted two years ago.

On the same occasion, the 27 EU Member States and MEPs also agreed—following a proposal by the European Commission—to postpone the entry into force of new rules governing so-called high-risk AI systems, meaning those operating in sensitive areas such as security, health, and fundamental rights.

These rules had originally been scheduled to take effect in August, but the Commission argued that businesses needed more time to adapt to this unprecedented regulatory framework.

The main change introduced by Member States and MEPs to the Commission’s proposal was the establishment of fixed implementation dates, rather than the more flexible timetable initially suggested by Brussels.

The new deadlines are 2 December 2027 for standalone high-risk systems, and 2 August 2028 for those embedded in other systems or products.

The agreement comes amid renewed concerns in the EU over the risks associated with artificial intelligence, following recent attention on Mythos, a new model from US startup Anthropic, whose alleged cybersecurity capabilities have raised concerns.