The EU AI Act’s August 2, 2026 compliance deadline, the date high-risk AI providers and deployers have been planning around since the Act took effect, may be moving. Trialogue negotiations over a “Digital Omnibus” package are ongoing as of April 2026, and according to legal analysts tracking the negotiations, the package proposes delaying standalone high-risk AI system compliance to December 2027. A second track would push compliance for AI embedded in regulated products further still, to August 2028, according to analysts monitoring the process.
Neither delay is enacted. The Digital Omnibus package remains under active trialogue negotiation between the European Commission, Parliament, and Council. Until an agreement is reached and formally adopted, August 2, 2026 remains the operative deadline. Compliance teams planning around that date should not treat this proposal as a decision.
The part of this story that matters most for compliance professionals: legal analysts note that the proposed delays would not alter the substantive requirements under Articles 9 through 17 or the associated penalty structure. The documentation requirements, risk management obligations, and conformity assessment frameworks stay in place. An 18-month extension on the deadline is not an 18-month extension on the work.
There’s a second dimension worth tracking. Per Epoch AI’s compute tracking, the number of AI models meeting or exceeding the EU AI Act’s systemic risk threshold – set at 10^25 floating point operations, has grown significantly since the Act’s passage. The scope of who this regulation covers is expanding even as the timeline remains uncertain. Organizations that concluded they fell below the systemic risk threshold during the Act’s initial drafting may need to revisit that assessment.
The hub previously covered what Article 6 requires from compliance teams ahead of the August deadline, and how the EU AI Office clarified high-risk classification for agentic tools and HR systems. Today’s development directly updates the timeline context in both of those pieces.
What to watch: the trialogue’s next formal session, any trilateral agreement announcement from the European Commission, and whether the EU AI Office issues interim guidance acknowledging the negotiation’s status. If the Omnibus proposal advances, compliance teams will need updated guidance on which deadline applies to which systems, the two-track structure (standalone high-risk vs. AI embedded in regulated products) creates different timelines for different parts of the same organization’s AI portfolio.
The practical position for compliance teams right now is straightforward, even if the regulatory calendar isn’t: preparations built for August 2026 don’t expire in December 2027. Risk management documentation, data governance structures, and conformity assessment work have organizational value regardless of when the regulator formally requires them. The teams that will have the hardest time if the deadline holds are the ones who used this proposal as a reason to pause.