The EU AI Act's "Delay" Didn't Delay Everything
"The EU delayed the AI Act" was the headline in May, and a lot of people stopped reading right there. Only part of it actually moved. On August 2, 2026 — the date everyone thought got pushed back — a separate set of rules on labeling AI content and disclosing chatbots took effect exactly on schedule. Here's what the Digital Omnibus actually changed, what it left alone, and what that means if your tool or content reaches users in the EU.
The Digital Omnibus amends the AI Act's timeline without changing its underlying structure.
Some of the AI Act Moved. A Big Chunk Didn't.
The Digital Omnibus on AI, formally Regulation (EU) 2026/1744, entered into force on July 27, 2026. Its most significant change deferred the AI Act's high-risk system obligations under Annex III — the rules covering AI used in employment, education, credit scoring, law enforcement, critical infrastructure, and biometrics — from their original August 2, 2026 deadline to December 2, 2027. High-risk obligations for AI embedded in products already regulated under EU product-safety law, like medical devices or machinery, move further out, to August 2, 2028.
What the Omnibus did not touch: Article 50's transparency obligations, which took effect exactly on the original August 2, 2026 schedule. Those cover disclosing when someone is interacting with an AI system, labeling AI-generated or manipulated content, and flagging emotion-recognition or biometric-categorization use. General-purpose AI provider obligations, in force since August 2025, and the Article 5 prohibited-practices rules, in force since February 2025, were also left in place.
How the Deadline Actually Moved
November 19, 2025 — the Omnibus is proposed
The European Commission publishes the Digital Omnibus as part of a broader package touching the AI Act, GDPR, ePrivacy, NIS2, and the Data Act, citing the fact that the technical standards needed to make high-risk compliance operable weren't ready in time.
May 7, 2026 — Parliament and Council reach agreement
A provisional political agreement lands on postponing the high-risk deadlines, alongside new prohibitions on AI-generated non-consensual intimate imagery and child sexual abuse material.
June 16 & 29, 2026 — final adoption
Parliament endorses the final text on June 16; the Council gives final approval on June 29, clearing the way for publication.
July 27, 2026 — the Omnibus enters into force
The deferral becomes binding law, just six days ahead of the original August 2 high-risk deadline.
August 2, 2026 — the date arrives, and half the rules apply anyway
High-risk Annex III obligations don't take effect, as planned under the new schedule. Article 50 transparency and labeling obligations do take effect, exactly as originally scheduled, along with expanded AI Office enforcement powers.
Why "Delayed" Is the Wrong One-Word Summary
Most headlines in May and June led with "the EU delayed the AI Act," which is true but incomplete in a way that matters. The Act phases in several distinct categories of obligation on different timelines, and the Omnibus only touched one of them.
The practical distinction is between systems that make or influence significant decisions about people — hiring, credit, law enforcement, access to education — and systems that generate or present content to people. The first category, high-risk under Annex III, got the multi-year reprieve. The second category, transparency under Article 50, did not.
The Rules Live Since August 2, 2026
AI System Disclosure
Users must be told when they're interacting with an AI system rather than a human, in contexts where that wouldn't otherwise be obvious.
AI-Generated Content Labeling
Synthetic or manipulated image, audio, video, and text content must be marked in a way that's detectable as AI-generated, including deepfakes.
Emotion & Biometric Disclosure
Deploying emotion-recognition or biometric-categorization systems requires informing the people subject to them.
GPAI & Prohibited Practices
General-purpose AI provider duties (since August 2025) and the Article 5 ban on manipulative or exploitative AI practices (since February 2025) continue unchanged.
You Don't Have to Be Based in the EU
The AI Act's transparency obligations apply based on where an AI system is used, not where the company building it is headquartered. A chatbot, image generator, or content tool built anywhere in the world can fall under Article 50 the moment it's reaching users inside the EU.
For most tools and content operations, the practical checklist is short: disclose AI involvement where it isn't obvious, label synthetic media output, and don't build systems that rely on emotion or biometric inference without telling people. The high-risk deferral is genuine breathing room for anyone building in the Annex III categories — but it isn't a reason to skip the transparency work, which is already live and already enforceable.
Bottom Line
The EU's Digital Omnibus on AI, in force since July 27, 2026, pushed the AI Act's high-risk system obligations from August 2, 2026 to December 2, 2027 (and to 2028 for embedded systems) — a real and significant deferral. What it left untouched is Article 50: the rules requiring AI disclosure and content labeling, which took effect exactly on schedule on August 2, 2026, with enforcement powers and fines attached. If your work involves generating content or building conversational tools that reach EU users, the "delay" story doesn't apply to you — check your transparency obligations now.
More AI Guides, Written Clearly
Browse our full library of AI tool guides, updated regularly as models and regulations change.
Common Questions
Did the EU delay the AI Act?
Partly. The Digital Omnibus on AI, which entered into force on July 27, 2026, deferred the AI Act's high-risk system obligations (Annex III) from August 2, 2026 to December 2, 2027, and embedded high-risk obligations (Annex I) to August 2, 2028. It did not delay the Article 50 transparency and AI-content-labeling obligations, which took effect on schedule on August 2, 2026.
What is still required starting August 2, 2026?
Article 50 transparency duties: telling users when they're interacting with an AI system, labeling AI-generated or manipulated content (including deepfakes), and disclosing emotion-recognition or biometric-categorization use. General-purpose AI provider obligations, which have applied since August 2025, and the Article 5 prohibited-practices rules, in force since February 2025, also continue unaffected.
What obligations were pushed to 2027 and 2028?
High-risk system obligations under Annex III (covering employment, education, credit scoring, law enforcement, critical infrastructure, and biometrics) move to December 2, 2027. High-risk obligations for AI embedded in products already regulated under EU product-safety law, like medical devices or machinery, move to August 2, 2028.
Does this apply to businesses outside the EU?
Yes, if your product or service reaches users in the EU. The AI Act's transparency obligations apply based on where the AI system is used, not where the provider is headquartered, so a chatbot or content tool built outside the EU can still fall under Article 50 if EU users interact with it.
More from the creator
Explore two other projects from the same creator.