What's enforceable now. What's coming August 2, 2026.
Every date that matters, updated for the Digital Omnibus proposals. The one date that isn't moving: Article 50 transparency — August 2, 2026.
Visual timeline
Act enters into force
The EU AI Act was published in the Official Journal on July 12, 2024 and entered into force 20 days later. All transition periods start from this date.
What to do: Audit your AI use for prohibited practices. Implement basic AI training for staff.
Prohibited practices + AI literacy
All 8 prohibited AI practices are banned. AI literacy obligation (Article 4) requires all providers and deployers to ensure staff understand AI systems they use.
What to do: Audit your AI use for prohibited practices. Implement basic AI training for staff.
GPAI obligations + governance
General-Purpose AI model obligations apply (Articles 51-56). The AI Office, AI Board, and Scientific Panel are operational. Member states must designate national competent authorities. The GPAI Code of Practice (published July 10, 2025) is endorsed. 26 companies signed including Anthropic, Google, Microsoft, OpenAI, Amazon, and Mistral AI.
What to do: If you use GPAI models via API, verify your model provider has published their training data summary and technical documentation.
Article 50 transparency + GPAI fines + high-risk (Annex III)*
Article 50 transparency obligations are enforceable — this is confirmed and not subject to the Omnibus delay. GPAI fines can be imposed by the AI Office. Full enforcement powers for national authorities. *High-risk Annex III rules may be delayed to December 2, 2027 under the Digital Omnibus.
What to do: Implement Article 50 transparency. Complete your AI inventory. Prepare FRIA if required.
High-risk AI (Annex III) — delayed deadline
The Digital Omnibus on AI proposes pushing Annex III high-risk obligations to this date. European Parliament voted 101-9 in favour on March 18, 2026. Trilogue ongoing; adoption expected mid-2026.
What to do: Start compliance work now regardless — being ready early is a competitive advantage.
High-risk AI (Annex I) — product safety
AI that is a safety component of products covered by EU harmonisation legislation (medical devices, machinery, toys, etc.). Also the grandfathering deadline for GPAI models placed on market before August 2025.
Large-scale IT systems
Systems used in freedom, security, and justice areas (SIS, VIS, ETIAS, etc.).
The Omnibus — what's happening
On November 19, 2025, the European Commission published the "Digital Omnibus on AI" (COM(2025)0836), proposing to link high-risk deadlines to the availability of harmonised standards. The European Parliament's IMCO/LIBE committees rejected the conditional approach and instead set fixed dates:
| Category | Proposed Date |
|---|---|
| Annex III high-risk | December 2, 2027 |
| Annex I product safety | August 2, 2028 |
| Article 50(2) watermarking for pre-existing systems | November 2, 2026 |
Trilogue negotiations between Parliament, Council, and Commission are expected through spring 2026. The EU Council (Cyprus Presidency) broadly supports postponement. Until formally adopted, the original August 2, 2026 deadline remains legally binding.
Standards — the missing piece
No harmonised standards have been published. CEN-CENELEC JTC 21 adopted exceptional acceleration measures in October 2025, but the first standards (starting with quality management) aren't expected before Q4 2026. Over 400 reconsideration requests are pending on the risk management standard alone. This standards gap is the primary justification for the Omnibus delay.
National implementation
Finland: First EU member with fully operational enforcement (January 1, 2026, led by Traficom).
Spain: AESIA — first dedicated AI supervisory body.
Italy: National AI fines up to €774,685 and criminal penalties for deepfakes (Law No. 132/2025).
Denmark: Implementation legislation passed May 2025.
Germany, France, Netherlands: Partial implementation. At least 12 member states missed the August 2025 deadline to appoint competent authorities.
Deadline-aware classification
KlarvoEngine tracks all enforcement dates and generates obligations with the correct deadlines for your specific classification.
Classify your systems →Classify your AI systems. Free.
KlarvoEngine — 3-pass regulatory classification with article-cited memos. No credit card required.
Start your compliance journey