Overview
- Transparency duties became EU-wide law on 2 August 2026, requiring chatbots, voice assistants, avatars and similar systems to tell users they are interacting with AI.
- Platforms must add clear, readable labels to content that is generated or modified by AI, including deepfakes, while some systems already on the market face phased marking deadlines.
- The European AI Office will work with national authorities to take complaints and enforce the rules, with specific fines up to 3% of global turnover or €15 million for transparency breaches.
- More serious violations carry higher penalties ranging from €7.5 million to €35 million or up to 7% of global turnover, and providers of advanced general-purpose models face extra duties to prevent large-scale harms.
- Brussels pairs the law with a voluntary transparency code signed by more than 180 companies and a staged timetable that adds bans and high-risk rules from December 2026 through August 2028, a sequence that will test enforcement and industry compliance.