Europe's landmark AI rulebook just got a major rewrite weeks before its most demanding provisions were due to kick in. With the Council of the EU giving its final approval on June 29 to the so-called Digital Omnibus simplification package, the AI Act's high-risk obligations have been pushed back by more than a year — but the transparency rules arriving on August 2, 2026 remain firmly on the calendar, and so do the European Commission's new enforcement powers.
What the Digital Omnibus Changes
The Digital Omnibus package cleared its last two hurdles in quick succession. The European Parliament endorsed the amendments on June 16 by a vote of 423 to 57, with 174 abstentions, and the Council's sign-off on June 29 completed the legislative process.
The headline change is timing. Obligations for standalone high-risk AI systems listed in Annex III — covering areas such as hiring, credit scoring, education, and access to essential services — move from August 2, 2026 to December 2, 2027. High-risk AI embedded in regulated products, such as medical devices and machinery, shifts from August 2027 to August 2, 2028.
Lawmakers paired the delay with a few substantive additions:
- A new prohibition on AI tools that generate non-consensual intimate imagery or child sexual abuse material, aimed squarely at so-called "nudify" apps, applying from December 2, 2026.
- An extension of the Act's SME-friendly simplifications to mid-cap companies with up to 750 employees and €150 million in annual revenue.
- Clarifications making it easier to process sensitive personal data when the purpose is detecting and correcting bias in AI models.
Brussels has framed the package as a response to competitiveness concerns and to complaints that technical standards and guidance were not ready in time for companies to comply.
What Still Happens on August 2
Companies tempted to relax should read the fine print. The Article 50 transparency obligations were not postponed. From August 2, 2026, providers and deployers across the EU market must ensure that users are informed when they are interacting with a chatbot, that AI-generated content carries machine-readable markings, and that deepfakes are clearly labeled. The only cushion is for systems already on the market, which get until December 2, 2026 to implement machine-readable marking.
August 2 also marks the moment the Commission's enforcement toolkit for general-purpose AI models becomes fully operational. The substantive obligations for model providers have applied since August 2025, but from next month the AI Office can compel information, request model access, and ultimately push for restrictions or recalls. The Act's penalty ceilings — up to €35 million or 7 percent of global annual turnover for the most serious violations — remain untouched.
Extension, Not Absolution
Legal advisers across the bloc are delivering a consistent message: this is a deferral of deadlines, not a dilution of substance. The core high-risk requirements — risk management systems, data governance, technical documentation, human oversight, and conformity assessments — survive the Omnibus intact. Organizations that pause their compliance programs now will face the same mountain in December 2027, likely with less goodwill from regulators.
There is also a practical wrinkle for multinationals. Firms selling into both the EU and other jurisdictions are increasingly building a single global compliance baseline, and several U.S. states are moving ahead with their own AI statutes on independent timelines. A European delay does not slow those clocks.
Why It Matters
The Digital Omnibus is the clearest signal yet that Europe is recalibrating its approach to AI governance under competitive pressure from the United States and China. For AI builders, the extra time is genuinely valuable: harmonized standards, guidance, and conformity assessment infrastructure were lagging badly, and a rushed August rollout risked chaotic, uneven enforcement.
But the deadline that survived may matter more than the ones that moved. Transparency duties — telling people they are talking to a machine, watermarking synthetic media, labeling deepfakes — will reshape how AI products look and feel for hundreds of millions of Europeans starting next month. And with the Commission's general-purpose AI enforcement powers going live at the same moment, the world's most ambitious AI law is about to gain real teeth, even in its slimmed-down form. Businesses that treat the delay as a reprieve rather than a runway are misreading Brussels.
