Digital Omnibus Adopted: AI Act High-Risk Deadline Now December 2027

Yes. The Digital Omnibus on AI has been adopted as Regulation (EU) 2026/1744, published in the Official Journal on July 24, 2026 and in force since July 27, 2026. For companies subject to the AI Act (Regulation EU 2024/1689), obligations for high-risk AI systems listed in Annex III now apply from December 2, 2027 instead of August 2, 2026, and those for systems embedded in Annex I products from August 2, 2028. The Article 50 transparency obligations, however, have applied since August 2, 2026, as originally planned.

The text stems from the Commission's proposal of November 19, 2025, part of a simplification package that also covered the GDPR, the Data Act and other digital texts. It is no longer a proposal: these dates are now binding law.

Why the Commission Proposed This Delay

The Commission's reasoning was pragmatic: the compliance ecosystem was not ready. The harmonised standards needed to guide companies — developed by the CEN-CENELEC Joint Technical Committee 21 — were not expected before late 2026. Without these standards, companies cannot know with certainty whether their technical documentation is compliant. Imposing an obligation whose implementation tools do not yet exist creates legal uncertainty.

The Commission also identified structural delays in the designation of national competent authorities and conformity assessment bodies across Member States. The delay is therefore not a regulatory retreat, but what the Commission itself called a "structural recalibration".

What the Digital Omnibus Changes in Practice

New application dates for high-risk systems

The Commission's initial proposal linked the application of high-risk obligations to a Commission decision confirming that compliance support tools (standards, guidelines, common specifications) were available, with backstop dates. Since the adoption of Regulation (EU) 2026/1744, the dates to plan around are the following:

• December 2, 2027 for Annex III systems (e.g., AI in recruitment, education, essential services) — instead of August 2, 2026
• August 2, 2028 for Annex I systems (embedded in regulated products: medical devices, machinery, vehicles) — instead of August 2, 2027

In the amended calendar, national regulatory sandboxes (Article 57) must be operational by August 2, 2027.

Relief for SMEs and small mid-caps

The Commission's proposal introduced a new category: "small mid-caps", defined as companies with fewer than 750 employees and less than €150 million in annual turnover, intended to benefit from the same reliefs as classic SMEs. Refer to the text published in the Official Journal for the exact scope of these reliefs, which concerned in particular:

• Simplified technical documentation for high-risk systems
• Lighter quality management system (QMS) requirements
• Adapted fine calculation methods based on company size

Article 50: grace period for marking until December 2, 2026

Article 50 has applied since August 2, 2026. Only the machine-readable marking of AI-generated content (Article 50(2) — watermarks, metadata) benefits from a grace period until December 2, 2026, and only for systems placed on the market before August 2, 2026. Systems placed on the market from August 2, 2026 onwards must comply right away. To help implement these obligations, the Commission published its final guidelines on Article 50 on July 20, 2026, and a voluntary code of practice on marking and labelling AI-generated content on June 10, 2026.

AI literacy relaxed, new prohibitions

The Omnibus reformulates Article 4: AI literacy becomes an obligation to take measures for the AI literacy of staff — a more flexible wording than the original text. It also adds new prohibited practices, applicable from December 2, 2026: systems designed to generate non-consensual intimate images or child sexual abuse material.

What Doesn't Change

The Digital Omnibus does not alter the fundamental structure of the AI Act. The following remain unchanged:

• Prohibited practices (Article 5): enforceable since February 2, 2025 — the existing prohibitions are not affected
• GPAI obligations (Articles 51–56): in force since August 2, 2025
• Risk classification: Annex III, risk levels, criteria — nothing changes
• Article 50 transparency obligations: applicable since August 2, 2026
• Penalty levels: up to €35M or 7% of global turnover for prohibited practices, €15M or 3% for other obligations

For high-risk systems, the Digital Omnibus does not remove the requirements — it delays their enforcement. The Annex IV technical documentation, risk management (Art. 9), data governance (Art. 10): these obligations still exist. Only their enforceability date changes.

From Proposal to Regulation (EU) 2026/1744: the Timeline

The Digital Omnibus on AI followed the ordinary legislative procedure, in about eight months:

• November 19, 2025: Commission proposal (COM(2025) 836)
• May 7, 2026: provisional agreement in trilogue; June 16, 2026: final vote in the European Parliament; June 29, 2026: adoption by the Council
• July 24, 2026: publication in the Official Journal, then entry into force on July 27, 2026

The text was therefore adopted before the original August 2, 2026 deadline for high-risk systems, which no longer applies to Annex III systems. The legislative debate is closed: the dates above are now the ones that bind companies.

What Should You Do Now?

Now that the deadline has moved, the temptation to slow down compliance efforts is understandable. It would, however, be a strategic mistake for three reasons:

1. The new dates are law. December 2, 2027 (Annex III) and August 2, 2028 (Annex I) are set in Regulation (EU) 2026/1744. There is no longer a legislative outcome to wait for.
2. The delay gives months, not years. Technical documentation, risk management and quality management for a high-risk system take many months to build. Companies that start compliance now will be far better positioned than those who wait.
3. Risk classification doesn't change. Knowing whether your system is high-risk is a prerequisite for any compliance effort — and it doesn't depend on any harmonised standard.

The safest strategy: comply with Article 50 now, and plan your high-risk compliance backwards from December 2, 2027. The free AiActo diagnostic lets you classify your AI systems and identify your obligations right now — whatever their risk category.

Frequently Asked Questions

Has the Digital Omnibus been adopted?

Yes. The Digital Omnibus on AI was adopted as Regulation (EU) 2026/1744, after a provisional trilogue agreement on May 7, 2026, a European Parliament vote on June 16, 2026 and adoption by the Council on June 29, 2026. It was published in the Official Journal on July 24, 2026 and has been in force since July 27, 2026.

Do the August 2026 AI Act obligations still apply?

Partly. The Article 50 transparency obligations have applied since August 2, 2026, unchanged by the Omnibus (with a grace period until December 2, 2026 for the marking of content generated by systems already on the market before that date). High-risk obligations under Annex III, on the other hand, now apply from December 2, 2027.

Which companies benefit from the new SME exemptions?

The Commission's proposal extended reliefs to classic SMEs (fewer than 250 employees, less than €50M turnover) and to a new category of "small mid-caps" (fewer than 750 employees and less than €150M in annual turnover), with simplified technical documentation and quality management requirements. Check Regulation (EU) 2026/1744 as published for the exact scope of these measures.

Are GPAI obligations affected by the delay?

No. Obligations for providers of general-purpose AI models (Articles 51 to 56) have been in force since August 2, 2025 and are not postponed by the Digital Omnibus.

What changes for AI literacy (Article 4)?

The AI literacy obligation has applied since February 2, 2025. The Omnibus reformulates it: it is now an obligation to take measures for the AI literacy of staff, a more flexible wording than the original text. Putting such measures in place therefore remains necessary.

The Digital Omnibus does not change the substance of the AI Act — it adjusted its enforcement timeline to reflect a concrete reality: the compliance tools were not ready. But the delay does not change the direction. Classifying your AI systems, inventorying your obligations and structuring your documentation remain unavoidable steps, whatever their risk category. Check the AI Act timeline on AiActo to follow regulatory calendar updates in real time.