The EU AI Act: what actually changed in May 2026

On 7 May 2026, the EU agreed the Digital Omnibus on AI — a package that reshuffled the AI Act's timeline. Most of the coverage got it half-right. Here is what it actually means for companies using AI in their day-to-day operations.

Postponed: the heavy high-risk obligations (Annex III) move from 2 August 2026 to 2 December 2027, and AI embedded in regulated products (Annex I) gets until 2 August 2028. If you were racing toward an August high-risk deadline, you now have breathing room.

Not postponed: the Article 50 transparency rules still take effect on 2 August 2026. If you run a customer-facing chatbot or AI assistant, users must be told they are interacting with AI — and AI-generated content (text, images, audio) must be machine-readably marked. Systems already live before 2 August get a grace period on watermarking until 2 December 2026.

Good news for mid-size companies: the SME simplification regime was extended to companies with up to 750 employees and €150 million in revenue — simplified documentation, reduced fines, and regulatory sandbox access.

What to do now: inventory your AI touchpoints, check each one against the Article 50 duties (disclosure, content marking, deepfake labelling), and close the gaps before 2 August. It is a small, well-scoped exercise — and far cheaper before the deadline than after.

Want a second pair of eyes? Cegona runs a fixed-price, two-week EU AI Act Transparency Check. Book a free 30-minute assessment at cal.eu/axelsalomonsson or write to axel@cegona.com.

Published June 2026. This is general information, not legal advice.