EU member states and European Parliament negotiators have agreed to ban artificial intelligence applications used to create non-consensual sexualized deepfakes, the Cypriot EU Council presidency said early on Thursday.
At the same time, other rules regarding artificial intelligence (AI) are set to be simplified in order to unlock the economic potential of the technology in Europe.
Before the changes can take effect, the agreement still needs formal approval from the European Parliament's plenary session and EU member states. This is typically considered a formality.
If the measure is approved, companies will have until December 2, 2026 to bring their systems in line.
In deepfakes, a person's face can be inserted into another video, for example a pornographic sequence, or their voice can be artificially replicated, creating the impression that they are doing or saying things that were not done or said. Advances in artificial intelligence have made it significantly easier to produce highly convincing artificial content.
The planned ban is also explicitly intended to cover the creation of material depicting child sexual abuse.
"We are stepping up the protection of children targeting risks linked to the AI systems," said Marilena Raouna, deputy minister for European affairs of Cyprus, in a statement.
The non-consensual sharing of intimate or manipulated material, including deepfakes, was classified as a criminal form of cyber violence in the EU in 2024.