Two and a half weeks ago I wrote about the AI Act deadline shuffle and warned that planning around a provisional delay is a bet, not a strategy. The bet has now been settled. On 16 June the European Parliament endorsed the Digital Omnibus package, and on 29 June the Council gave its final green light[1][2]. The postponement of the Act’s high-risk obligations is no longer a proposal. It is adopted law awaiting publication in the Official Journal. Which means the real question has changed. It is no longer what applies when. It is what you do with the sixteen months you were just handed.
What is now final
The package moves the obligations for stand-alone high-risk systems under Annex III from 2 August 2026 to 2 December 2027, and for AI embedded in regulated products under Annex I from August 2027 to 2 August 2028[2][3]. It also adds a new prohibition to Article 5: systems built to generate non-consensual intimate imagery or child sexual abuse material: effective 2 December 2026[3]. What did not move matters more for the next two weeks: the Article 50 transparency duties: chatbot disclosure, AI-content marking, deepfake labelling: apply from 2 August 2026 as originally scheduled[2]. And the AI literacy duty in Article 4 has been in force since February 2025. It covers everyone whose people work with AI, which by now is everyone[1].
A window is not a pause
Here is the uncomfortable arithmetic behind the relief. Deloitte’s survey of 3,235 enterprise leaders finds close to three quarters of companies planning to deploy agentic AI within two years, while only 21 percent have a mature model for governing what those agents do[4]. That gap between deployment ambition and governance maturity does not close by itself, and it does not care about Brussels dates.
Sixteen months is roughly what it takes a mid-sized organisation to do this properly once: build the AI inventory, classify honestly, assign ownership, fold literacy into normal enablement and let the governance rhythm settle. Companies that start now walk into December 2027 with compliance as a by-product of good operations, and a trust signal their customers can check. Companies that read the delay as a pause will do the same work later, compressed, under pressure, at a worse price.
What I see in the field
Since the Omnibus vote I have heard the same sentence in a DAX headquarters environment and from mid-sized owners alike: good, so this is a 2027 problem. It is usually said in the same week a new AI tool gets approved. That is the pattern to watch: adoption accelerating and accountability postponed, in the same meeting. The teams I rate highest do something quietly different. They treat the new dates as a project deadline, not a relief: the inventory on one page, the risk class next to each system, one name next to the whole thing. None of them needed more than a quarter.
What this means for you
Three moves. First, take the next two weeks seriously: if you run customer-facing chatbots or publish AI-generated content, the Article 50 transparency duties arrive on 2 August 2026, that is current work, not future work[2]. Second, build the inventory and classify honestly; most of what you run is probably not high-risk, and knowing that reduces fear as much as risk. Third, plan backwards from December 2027 the way you would from any fixed milestone: ownership this quarter, governance rhythm next quarter, evidence trail from then on.
Mapping your AI Act exposure, which systems, which obligations, which owner: is one of the four dimensions of our AI Readiness Audit, and the Article 4 literacy duty is built into every AI Foundations Workshop, documented so you can show it happened. If your organisation just filed the AI Act under 2027, it may be worth one conversation before the quarter is gone.
Personalised answer within 48 hours.
Sources
- European Commission, AI Act: Regulatory framework for AI, implementation status. digital-strategy.ec.europa.eu
- Gibson Dunn, EU AI Act Omnibus Agreement: Postponed High-Risk Deadlines and Other Key Changes, 2026. gibsondunn.com
- Morgan Lewis, EU Approves Delays and Other Amendments to Certain EU AI Act Obligations, June 2026. morganlewis.com
- Deloitte, The State of AI in the Enterprise 2026 (survey of 3,235 leaders). deloitte.com
Briefed Is Not Fluent: The New Leadership Gap in AI