Gallery

Contacts

411 University St, Seattle, USA

engitech@oceanthemes.net

+1 -800-456-478-23

Skip to content
Regulation Daily Brief

EU Institutions Reach Provisional Agreement on Digital Omnibus, High-Risk Deadlines Extended, New AI Ban Added

3 min read artificialintelligenceact.eu Partial Strong
EU Parliament, Council, and Commission reached a provisional political agreement on the Digital Omnibus on May 27, 2026, extending Annex III high-risk compliance deadlines by 16 months and adding a new prohibition on AI systems designed to generate non-consensual intimate imagery. The August 2, 2026 Article 50 transparency deadline is not affected and remains fixed.
Annex III extension, 16 months

Key Takeaways

  • EU institutions reached a provisional political agreement on the Digital Omnibus on May 27, 2026, not enacted law; Official Journal publication is still required.
  • Per the provisional agreement, the Annex III high-risk compliance deadline would extend from August 2, 2026 to December 2, 2027 (16 months).
  • The Article 50 transparency deadline, watermarking, labeling, chatbot disclosure, remains August 2, 2026 and is not affected by the Omnibus.
  • A new Article 5 prohibition on AI systems designed to generate non-consensual intimate imagery is included in the provisional agreement.

Timeline

2026-05-27 Provisional political agreement reached on Digital Omnibus
2026-08-02 Article 50 transparency deadline, CONFIRMED, unaffected
2027-12-02 Annex III high-risk deadline, per provisional agreement
2028-08-02 Annex I embedded safety systems deadline, per provisional agreement

Compliance Deadline

August 2, 2026
66 days remaining
EntityEU AI Office
JurisdictionEU
PenaltyUp to 1.5% of global annual turnover

EU institutions shook hands on the Digital Omnibus. What that means for your compliance program depends almost entirely on which deadline you’re watching, because the deal doesn’t move all of them.

What happened

On May 27, 2026, the European Parliament, Council, and Commission reached a provisional political agreement on the Digital Omnibus package, which includes amendments to the EU AI Act’s compliance timeline. Per the provisional agreement, the Annex III high-risk compliance deadline would move from August 2, 2026 to December 2, 2027, a 16-month extension. The Annex I deadline, covering AI embedded in regulated safety products such as medical devices and machinery, would move from August 2, 2027 to August 2, 2028, per legal analysis of the provisional text.

The agreement also includes a new Article 5 prohibition. Article 5 of the EU AI Act already prohibits specific high-risk AI practices. The Digital Omnibus adds an explicit ban on AI systems specifically intended to create non-consensual intimate imagery, a provision with direct implications for generative AI providers operating image and video generation products in EU markets.

Who This Affects

GenAI Providers & Deployers
Article 50 transparency obligations (watermarking, labeling, chatbot disclosure) are due August 2, 2026, no extension. Begin final compliance checks now.
Annex III High-Risk System Operators
The provisional agreement targets December 2, 2027, but it isn't enacted law yet. Maintain your compliance program until Official Journal publication confirms the extension.
Image & Video Generation Providers
The new Article 5 prohibition on AI systems designed to generate non-consensual intimate imagery applies. Audit product capabilities and use-case terms before the Omnibus is formally adopted.

What doesn’t move

The Article 50 transparency deadline stands. August 2, 2026 remains the enforcement date for watermarking requirements, AI-generated content labeling, and chatbot disclosure obligations, per the European Commission’s digital strategy guidance. No part of the Digital Omnibus touches Article 50. Providers and deployers of generative AI systems, chatbots, and synthetic media tools have roughly nine weeks from today to be compliant.

One nuance compliance teams should verify: per analysis of the provisional text by Latham & Watkins, the agreement reportedly includes a grandfathering provision for certain generative AI systems placed on the market before a specified cutoff date. The scope of that grandfathering is not confirmed in primary regulatory text, the Official Journal has not yet published the amended Act. Verify applicability with legal counsel before relying on it.

What this agreement actually is

“Provisional political agreement” is a specific procedural term. It means the three EU institutions have reached consensus at the political level. It does not mean the amendment is enacted law. The formal adoption votes and Official Journal publication still lie ahead. Until publication, the original EU AI Act text remains the operative legal instrument. Organizations that built compliance programs around the original August 2026 Annex III deadline can’t formally stand down until enactment is confirmed.

Verification

Partial T1 (artificialintelligenceact.eu, EC digital strategy) for Article 50 and Article 5 prohibitions; T2/T3 corroboration for Annex III and Annex I deadline extensions Official Journal has not published the amended text as of May 27, 2026. Provisional agreement is not enacted law. Deadline extensions are qualified, use 'per the provisional political agreement' framing.

What to watch

The critical variable is timing between now and August 2. If the Official Journal publishes the amended text before August 2, 2026, Annex III compliance teams get the full 16-month runway. If it doesn’t, the legal situation between political agreement and enacted law creates a brief period of ambiguity. The Article 50 deadline moves in neither scenario.

TJS synthesis

The catch is sequencing. The Digital Omnibus gives Annex III deployers breathing room, but only once it’s law, not when it’s a handshake. The real question is whether your organization has mapped which obligations fall under Article 50 versus Annex III, because those two tracks now have completely different timelines and completely different certainty levels. Organizations that haven’t separated those tracks in their compliance programs face a two-speed planning problem that no amount of political goodwill resolves for them.

View Source
More Regulation intelligence
View all Regulation

Related Coverage

Stay ahead on Regulation

Get verified AI intelligence delivered daily. No hype, no speculation, just what matters.

Explore the AI News Hub