Story perspectives

EU Approves AI Act Overhaul, Bans Sexual Abuse Content.

5/8/2026

Summary

EU Amends AI

The EU reached an agreement on 7 May 2026 to amend the AI Act.
Stand-alone high-risk AI systems apply from 2 December 2027 and embedded ones from 2 August 2028.
The deal bans AI-generated non-consensual sexual content and child sexual abuse material effective 2 December 2026.
Providers must watermark AI output by 2 December 2026, cutting the grace period to three months.
All AI systems must register in the EU AI database, even if exempt from high-risk rules.

Summary

EU Bans Deepfakes

EU ban on AI-generated non-consensual intimate imagery is hailed as protection for women and girls by Irish MEP Michael McNamara.
Internet Watch Foundation reported a more than 260-fold rise in AI-generated child sexual-abuse videos in 2025, prompting urgency for the ban.
EU officials said the AI Office will gain access to providers’ internal safety practices, including Anthropic’s Mythos model, to address cybersecurity threats.

History

Historical Context

The EU’s 1995 Data Protection Directive first unified privacy rules across member states, a framework later expanded by the GDPR (adopted 14 April 2016, effective 25 May 2018). Building on that legacy, the 2026 AI Act amendment now obliges providers such as Anthropic to embed watermarking and faces fines up to €20 million or 4 % of global turnover, overseen by the European Commission and the European Data Protection Supervisor.

1 of 3

EU Amends AI

The EU reached an agreement on 7 May 2026 to amend the AI Act.
Stand-alone high-risk AI systems apply from 2 December 2027 and embedded ones from 2 August 2028.
The deal bans AI-generated non-consensual sexual content and child sexual abuse material effective 2 December 2026.
Providers must watermark AI output by 2 December 2026, cutting the grace period to three months.
All AI systems must register in the EU AI database, even if exempt from high-risk rules.

1 / 3