EU bans sexual deepfakes as German lawmakers increasingly use AI for speeches

In 2026, a Focus magazine analysis using the AI‑detection tool Pangram identified that 9 % of Bundestag plenary speeches were fully generated by artificial intelligence. Of the 853 speeches analyzed, 76 were classified as AI‑written, with the AfD leading at 23 instances, followed by the SPD and Union with 20 each. The SPD’s health spokesperson, Christos Pantazis, had all six of his speeches that year produced by AI, emphasizing that the content remained his own. Other politicians, including members of the CDU, have also faced scrutiny for using AI in speeches and guest articles.

On 11 June 2026, the European Parliament adopted new AI‑ban provisions targeting sexualized deepfakes. Providers of image‑generating chat‑bots must prove, by 2 December 2026, that their systems cannot create intimate depictions of identifiable persons without consent. The rules, part of the EU’s AI Act package, place responsibility on tool developers rather than just on users, but technical implementation and definition of “adequate safeguards” remain challenging for the industry.