AI Act — high-risk systems
The bans and GPAI duties are already live. The bulk of high-risk product rules (hiring, credit, essential services) switch on this month.
People
If a bank, hospital, school or employer uses AI on you, they should be able to explain the system and keep a human responsible. Chatbots and deepfakes should already be labelled.
Business
Inventory, classify, and demand technical files from vendors. High-risk deployers and providers face documentation, logging, and human-oversight duties. Fines can reach a share of worldwide turnover.
Watch: National market-surveillance authorities starting real files, not just guidance PDFs.