EU AI Act: Application Status as of 2 August 2026
Dated: 2 August 2026 · Series: Institute Briefs · Status: informational; not legal advice.
Today is the EU AI Act’s general application date under Article 113 — and the most widely misreported milestone in the Act’s rollout. Much of today’s commentary will say the Act is now “in full effect.” It is not. This brief records what actually applies as of today, what moved, and the Institute’s position on what that means for governance practice.
What changed
On 24 July 2026, Regulation (EU) 2026/1744 — the “Digital Omnibus on AI” — was published in the Official Journal of the European Union. It entered into force on 27 July 2026, six days before the AI Act’s general application date, and is the first substantive amendment to Regulation (EU) 2024/1689 (the AI Act). Its most consequential change rewrites Article 113’s application schedule:
- Stand-alone high-risk AI systems (Annex III) — employment, education, credit assessment, law enforcement, critical infrastructure, and similar — now apply from 2 December 2027 (previously 2 August 2026).
- High-risk AI embedded in regulated products (Annex I) — medical devices, machinery, toys, aviation — now apply from 2 August 2028 (previously 2 August 2027).
- These are fixed dates, not the standards-availability trigger the Commission originally proposed. No mechanism remains to move them again without new legislation.
What applies as of today, 2 August 2026
The general application date itself did not move. Provisions not expressly deferred become applicable today, most notably:
- Article 50 transparency obligations. People interacting with an AI system must be informed they are doing so. Synthetic audio, image, video, and text content must be marked as artificially generated or manipulated; deepfakes must be disclosed. Providers of generative systems placed on the market before 2 August 2026 have until 2 December 2026 to comply with Article 50(2).
- Enforcement powers over general-purpose AI. GPAI model obligations have applied since 2 August 2025; the Commission’s powers to investigate and fine arrive today.
- The remainder of the non-deferred framework, layered on what was already in force: the Article 5 prohibitions and the Article 4 AI literacy duty (applicable since 2 February 2025 — the Omnibus recast literacy as an obligation of effort, with national supervision and enforcement beginning 2 August 2026), and the governance and penalties provisions applicable since 2 August 2025.
Consolidated timeline
| Date | What applies |
|---|---|
| 1 Aug 2024 | AI Act enters into force |
| 2 Feb 2025 | Article 5 prohibitions; Article 4 AI literacy |
| 2 Aug 2025 | GPAI model obligations; governance bodies; penalties framework |
| 27 Jul 2026 | Regulation (EU) 2026/1744 (Digital Omnibus on AI) in force |
| 2 Aug 2026 | General application of non-deferred provisions, incl. Article 50 transparency; Commission GPAI enforcement powers; national supervision of the AI-literacy duty |
| 2 Dec 2026 | New prohibitions: AI for non-consensual intimate imagery and CSAM; Article 50(2) deadline for generative systems placed on the market before 2 Aug 2026 |
| 2 Aug 2027 | National regulatory sandboxes operational |
| 2 Dec 2027 | High-risk obligations, Annex III (stand-alone systems) |
| 2 Aug 2028 | High-risk obligations, Annex I (product-embedded systems) |
The Institute’s position
Organizations that budgeted to be high-risk-ready today have just been handed sixteen months of slack. The governance question is what an AI-IRB does with it. Our answer: nothing changes. In the G-12 to G-0 framework, the gates are commitments to the people a system affects, not to a regulator’s calendar. G-0 — continuous oversight — is a standing state, not a compliance date. Deadlines moved; obligations did not.
Crosswalk: G-12→G-0 framework ↔ EU AI Act
Indicative mapping of the gate-stage themes as taught in Chapter 4 of the AI-IRB Governance Certification course to the Act’s principal corresponding provisions. The gate framework’s formal per-gate definitions await council ratification (see CLAIMS.md, entry 1); article obligations phase in per the timeline above.
| Gate-stage theme | AI Act provisions | Status on 2 Aug 2026 |
|---|---|---|
| Ethical foundations, values, fundamental-rights screening (G-12 region) | Art. 27 fundamental rights impact assessment | Deferred with high-risk (2 Dec 2027 / 2 Aug 2028) |
| Prohibited-practice screening | Art. 5 | Live since 2 Feb 2025; new NCII/CSAM prohibitions from 2 Dec 2026 |
| Risk assessment and management | Art. 9 | Deferred with high-risk |
| Data governance and quality | Art. 10 | Deferred with high-risk |
| Technical documentation and record-keeping | Arts. 11–12 | Deferred with high-risk |
| Transparency to affected people | Art. 50 | Live today |
| Human oversight design | Art. 14 | Deferred with high-risk |
| Accuracy, robustness, cybersecurity | Art. 15 | Deferred with high-risk |
| Workforce AI literacy | Art. 4 | Live since 2 Feb 2025; obligation of effort; national supervision from 2 Aug 2026 |
| Foundation-model diligence | Arts. 53, 55 | Obligations live since 2 Aug 2025; Commission enforcement powers from today |
| Continuous oversight, post-market monitoring, incident response (G-0) | Arts. 72–73 | Deferred with high-risk — the Institute’s G-0 does not wait |
Sources
- Regulation (EU) 2024/1689 (AI Act), Official Journal text
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), OJ L, 24 July 2026; in force 27 July 2026
- European Commission: AI talent, skills and literacy
- European Commission: Article 50 transparency guidelines
- European Commission: Guidelines for providers of general-purpose AI models
- Retrieved and verified 2 August 2026. Corrections to this brief, if any, will be made by dated follow-up briefs, not by rewriting this one.