2 August 2026 is widely treated as the EU AI Act deadline. It is not. It is one date in a phased calendar — some obligations start on it, some started in 2025, and a large block that used to sit on this date has moved to 2027 and 2028.
This matters because getting it wrong in either direction is expensive. Assume nothing applies and you can miss transparency duties that do land on this date. Assume everything applies and you spend a compliance budget in 2026 that you did not need to spend until 2027.
Here is the accurate version, updated for the Digital Omnibus on AI — adopted as Regulation (EU) 2026/1744 and in force since 27 July 2026.
A quick recap — what already applies before August 2026
- 2 February 2025 — prohibited AI practices (Art. 5). Social scoring, emotion recognition in workplaces and schools, untargeted facial-image scraping, manipulative or exploitative systems, and the other banned uses. Breaches sit in the top penalty tier.
- 2 February 2025 — AI literacy (Art. 4). Every deployer must take measures to support AI literacy among staff who use its systems. National supervision of this duty begins 3 August 2026.
- 2 August 2025 — general-purpose AI model obligations, governance structures, and national authority designation. This is what applies to model providers such as OpenAI, Anthropic, Google and Mistral, plus the national market surveillance and notifying authorities.
None of this was touched by the Omnibus. It is already law, already in force, and already enforceable.
What was scheduled to take effect on 2 August 2026 (original framework)
Under the original AI Act text, 2 August 2026 was the date the stand-alone high-risk regime under Annex III started applying — AI used in employment and worker management, education, credit scoring, insurance pricing, essential services, law enforcement and migration.
That meant risk management systems, data governance, technical documentation, logging, human oversight, accuracy and robustness testing, conformity assessment, CE marking, and registration in the EU database — all due on the same day, for providers and (in a lighter form) deployers.
What the Omnibus changed about this date
The Commission proposed the Digital Omnibus in November 2025 because the harmonised standards companies need in order to demonstrate high-risk conformity were not going to exist in time. Parliament and Council agreed provisionally in May 2026, and the final regulation entered into force on 27 July 2026.
One detail is worth knowing, because a lot of secondary coverage still gets it wrong: the Commission originally proposed conditional start dates tied to a standards-readiness decision. Parliament and Council rejected that and turned the backstops into fixed dates. You are planning against a hard calendar, not a floating one.
EU AI Act application dates after the Digital Omnibus on AI (Regulation (EU) 2026/1744).
| Date | What applies | Status |
|---|---|---|
| 2 February 2025 | Prohibited AI practices are banned (Art. 5) | Unchanged |
| 2 February 2025 | AI literacy obligation applies (supervision from 3 August 2026) (Art. 4) | Unchanged |
| 2 August 2025 | General-purpose AI model rules and governance apply (Art. 51–55) | Unchanged |
| 2 August 2026 | Transparency duties apply — not deferred by the Omnibus (Art. 50) | Unchanged |
| 2 December 2026 | Machine-readable marking of synthetic output for generative systems placed on the market before 2 August 2026 (Art. 50(2)) | New |
| 2 December 2026 | Two new prohibitions apply (non-consensual intimate imagery, AI-generated CSAM) (Art. 5) | New |
| 2 August 2027 | National AI regulatory sandboxes must be operational (Art. 57)Previously 2 August 2026 | Deferred |
| 2 December 2027 | Stand-alone high-risk AI obligations (Annex III) apply (Annex III)Previously 2 August 2026 | Deferred |
| 2 August 2028 | High-risk AI embedded in regulated products (Annex I) applies (Annex I)Previously 2 August 2027 | Deferred |
So what actually happens on 2 August 2026?
Three things:
- Article 50 transparency duties apply. Tell people when they are interacting with an AI system. Label deepfakes and AI-generated content published on matters of public interest. Mark synthetic output in machine-readable form. Read Article 50 in practice.
- The AI Office gains full enforcement powers over general-purpose AI model providers.
- National supervision of the AI literacy duty begins (3 August 2026).
What does not happen: the high-risk regime does not start. Annex III systems apply from 2 December 2027, and high-risk AI embedded in regulated products from 2 August 2028.
The practical framing: treat 2 August 2026 as a checkpoint, not a finish line. It is the moment to confirm your system inventory, your role for each system, and your risk classification are in order — so the 2027 date is a documentation exercise rather than a rebuild.
What SMEs should actually do before this date
- Inventory every AI system you build or use. Name, owner, purpose, data used, who it affects. Shadow AI in marketing and HR counts.
- Confirm your role per system. Provider, deployer, importer or distributor. The obligations differ sharply — see do I need to comply?
- Classify risk per system. Most SME systems land in minimal or limited risk. Check where yours sits with risk classification explained.
- Fix your transparency surface now. AI disclosure in chat interfaces, labelling of AI-generated content, machine-readable marking of synthetic output. This is the part that genuinely lands in August 2026.
- Check SME and small mid-cap relief eligibility. Simplified technical documentation, proportionate quality management, lower fine caps and priority sandbox access are now written into the Act.
FAQ
Is 2 August 2026 still a real deadline?
Yes — for transparency, GPAI enforcement and AI literacy supervision. No — for the high-risk regime, which moved to 2 December 2027 (Annex III) and 2 August 2028 (Annex I regulated products).
Could the new dates move again?
They are fixed dates in an adopted regulation, not conditional triggers, so they will not drift on their own. A further amendment would take another full legislative process. Plan against 2 December 2027.
Do the obligations that started in 2025 still apply after this date?
Yes. Prohibited practices, AI literacy and the general-purpose AI model rules are unaffected by the Omnibus and remain fully in force.
Where can I check if my company is affected?
Run the free scan below, or work through the AI Act compliance checklist first.
Where to go next
For the full picture of what the Omnibus changed, read the Omnibus explainer. For every date in one view, see the EU AI Act timeline. To turn this into concrete actions, use the compliance checklist.
Indicative guidance only — not legal advice. ActNavigator provides guidance based on the EU AI Act (Regulation (EU) 2024/1689) as amended by Regulation (EU) 2026/1744.