EU AI Act High-Risk Deadline Moved to Dec. 2, 2027 — Digital Omnibus on AI, Regulation (EU) 2026/1744

For most of 2024 and 2025, compliance calendars, vendor marketing and news coverage converged on a single headline date for the European Union's Artificial Intelligence Act: 2 August 2026, when the obligations for high-risk AI systems were scheduled to become applicable. That date is now wrong for the high-risk category, and a great deal of material published before mid-2026 has not been updated. The change came through Regulation (EU) 2026/1744, formally titled a regulation amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 as regards the simplification of the implementation of harmonised rules on artificial intelligence, and known in shorthand as the Digital Omnibus on AI. It was adopted on 8 July 2026, published in the Official Journal of the European Union on 24 July 2026, and entered into force on 27 July 2026 — the third day following publication, an accelerated schedule the legislature adopted specifically so the amendment would land before the original August deadline passed. The instrument amends the AI Act itself alongside the EASA aviation safety regulation and the Machinery Regulation, reflecting how deeply the AI Act's high-risk regime is wired into existing EU product-safety law rather than standing apart from it.^[1][3][4]

The deferral splits the high-risk category along the same seam the AI Act itself uses. Systems classified as high-risk under Article 6(2) and listed in Annex III are stand-alone systems whose risk arises from their use case rather than from being part of a regulated product. These include tools used in recruitment and employment decisions, credit scoring and access to essential services, education, law enforcement, migration and border control, and administration of justice. For this group, the main compliance obligations now apply from 2 December 2027 rather than 2 August 2026 — a shift of roughly sixteen months. The second group covers systems classified as high-risk under Article 6(1) and Annex I: AI that functions as a safety component of, or is itself, a product already subject to EU harmonisation legislation. Medical devices, machinery and motor vehicles are the standard examples. Those obligations now apply from 2 August 2028. The longer runway for Annex I reflects the need to align AI Act conformity assessment with the separate certification cycles those products already run through, which is also why the omnibus had to amend the machinery and aviation regulations directly. Organisations that had built remediation programmes against the August 2026 date have additional time, but the obligations themselves — risk management systems, data governance, technical documentation, human oversight, accuracy and robustness requirements, and conformity assessment — were not narrowed by the postponement of the date on which they bite.^[1][3][4]

The most consequential practical point for companies shipping generative AI products is that 2 August 2026 remained a live compliance date. The Article 50 transparency obligations were carved out of the deferral and began to apply on schedule. The European Commission summarises them in three parts. Providers must design AI systems so that individuals are informed when they are interacting with an AI system rather than a human — the chatbot disclosure duty. Providers must also apply a machine-readable mark to synthetic content that an AI system generates or manipulates, and must enable that mark to be detected, which is the provenance and watermarking requirement that has driven adoption of content credential standards. Separately, deployers must disclose deepfakes and must label AI-generated or AI-manipulated text published on matters of public interest, using clear and perceivable labels, subject to limited carve-outs for human-reviewed editorial content and artistic or satirical works. The Commission indicates a transitional grace period until December 2026 for the marking obligation as applied to generative AI systems that were already placed on the market before 2 August 2026, and states that deepfakes generated before that date carry no mandatory retroactive labelling requirement, though labelling is encouraged. Enforcement carries real exposure: the Commission's published figures put fines for transparency breaches at up to EUR 15 million or up to 3% of total worldwide annual turnover, whichever is higher, with a separate ceiling of EUR 750,000 for EU institutions, bodies and agencies and proportionality adjustments for small and medium-sized enterprises. Two other tranches of the AI Act likewise ran unaffected: the Article 5 prohibited-practices regime, in force since February 2025, and the general-purpose AI model provider obligations, applicable since August 2025.^[2][4][5]