The Council of the European Union has agreed to proposals that would prohibit artificial intelligence systems from generating non-consensual intimate content, including applications designed to digitally alter images to remove clothing.
The measure forms part of updates to the European Union’s Artificial Intelligence Act under a broader legislative package known as the Digital Omnibus. The changes are intended to align rules across artificial intelligence, cybersecurity, and data protection within the bloc.
According to the proposal, AI systems that create non-consensual sexual or intimate content would be explicitly banned. The same provision also prohibits the use of AI to generate child sexual abuse material. The Council stated that these practices would be classified as unacceptable uses of artificial intelligence under EU law.
The updated rules also include requirements for providers of certain AI systems. Companies developing high-risk AI tools must register their systems in a European database, allowing regulators to monitor compliance. The proposal further clarifies the responsibilities of the EU’s AI Office and national authorities in enforcing the rules.
In addition to restrictions on harmful uses, the proposal introduces changes to how personal data can be processed in AI systems. It reinstates a requirement that the use of sensitive data must meet a standard of strict necessity when applied to tasks such as bias detection and correction.
The legislative update also includes measures aimed at reducing regulatory burden for smaller companies. Certain small and medium-sized enterprises may be exempted from specific obligations under the AI Act, according to the European Commission’s proposal.
The move follows earlier concerns raised by lawmakers and regulators about the use of AI tools to generate manipulated images of individuals without consent. Previous discussions within the European Parliament highlighted that such applications could violate fundamental rights, particularly when used to create sexualized or explicit content involving real people.
The Council’s agreement marks a step in the legislative process. The updated provisions will be incorporated into the EU’s AI regulatory framework, which sets out categories of permitted and prohibited uses of artificial intelligence across member states.