EU Parliament Approves Ban on AI for Creating Sexualized Deepfakes

The European Parliament has overwhelmingly approved a ban on artificial intelligence applications capable of generating sexualized deepfakes. A significant majority of 423 lawmakers voted in favor of amending the European AI Act, with 57 against and 174 abstentions. The legislation now awaits formal approval from EU member states.

Enforcement of the ban will be led by the EU's AI Office, established two years ago, starting December 2, 2026. Deepfakes, which use AI to create highly realistic manipulated images, videos, or audio, will face stricter regulations. The new rules extend beyond targeting users to also encompass providers of the digital tools used to create such content.

“

The new prohibition is intended to supplement existing rules by targeting not only the users but also the providers of the relevant digital tools.

Lawmakers involved in the negotiations emphasized that the ban is not intended to unduly restrict the creation or manipulation of images. The legislation will precisely define what constitutes sexualized content, focusing on AI applications that enable the creation of realistic images or videos of intimate body parts, including genitals, anuses, buttocks, or exposed female breasts, as well as sexually explicit acts. Satirical caricatures are explicitly excluded from the ban.

Further amendments to the AI Act aim to streamline regulations for the AI industry and facilitate its application in the economy. Companies behind AI models like ChatGPT and Claude will be required to comply with certain provisions from December 2026, including the mandatory labeling of AI-generated content with watermarks on images and videos. The ban is also intended to target the creation of content depicting child sexual abuse.

“

The ban is not intended to lead to an excessive restriction on the creation or manipulation of images.