On July 13, 2026, the Council of the European Union surfaced a Draft Final Act reference for Regulation (EU) 2024/1689, the Artificial Intelligence Act, in its public register search (interinstitutional file 2024/0106, register reference 00024/2024/LEX). The underlying text is the final adopted act dated June 13, 2024, already published in the Official Journal on July 12, 2024 and in force across the European Union since August 1, 2024. The register entry is bookkeeping, not new law, but it lands three weeks before the AI Act's general-applicability date of August 2, 2026, the deadline reshaping EU life-sciences and MedTech compliance.
For medical-device and IVD manufacturers, the operative question is how the AI Act's high-risk regime layers on top of MDR (Regulation (EU) 2017/745) and IVDR (Regulation (EU) 2017/746) over the next 24 months. The Digital Omnibus provisional agreement of May 7, 2026 deferred the Annex I high-risk applicability date for AI-enabled medical devices and IVDs to August 2, 2028, while the general-applicability date of August 2, 2026 still bites for the rest of the AI Act's obligations.
What actually changed in the Council register on July 13, 2026?
The Council's register search for legislative acts under reference "24/24" now returns the Draft Final Act entry signed by the Presidents of the European Parliament and the Council. That text is identical to the Official Journal version of Regulation (EU) 2024/1689, and only the Official Journal version is legally binding. No new obligation attaches to the surfacing itself; the practical signal is timing. The update is a reminder, three weeks before August 2, 2026, that the general-applicability milestone is within a single budget cycle. Compliance teams that had parked AI Act readiness behind the long runway are out of runway.
Who is hit by the August 2, 2026 general-applicability date?
From August 2, 2026, the bulk of the AI Act's obligations become applicable, including the transparency duties for general-purpose AI models and the remaining provisions not already in force since February 2, 2025 (prohibited practices) and August 2, 2025 (GPAI and notifying authority designation). For life-sciences firms, AI systems used in drug discovery, regulatory operations, pharmacovigilance triage, clinical-trial recruitment and quality-management automation move from "monitor" to "must-comply" status, even when they do not yet qualify as high-risk medical devices.
AI systems that are safety components of regulated products on the market, or that are themselves regulated products awaiting a third-party conformity assessment under MDR, IVDR or similar harmonised legislation, are high-risk under Article 6(1) and carry the full obligation set: risk management, data governance, technical documentation, logging, human oversight, accuracy, robustness and cybersecurity. The Article 6(1) Annex I high-risk applicability date for medical devices and IVDs is the one that was deferred.
When do AI-enabled medical devices and IVDs have to be AI Act compliant?
Under the Digital Omnibus provisional agreement of May 7, 2026, the Article 6(1) Annex I high-risk applicability date for AI systems that are safety components of, or themselves are, medical devices or IVDs is pushed to August 2, 2028. The deferral is conditional on the Digital Omnibus being formally adopted before August 2, 2026; if formal adoption slips past that date, the original August 2, 2027 date reverts. MDCG 2025-6 guidance on the AI Act / MDR / IVDR interplay remains the operational reference for notified bodies preparing dual conformity assessments.
The deferral does not remove MDR or IVDR obligations: a CE-marked AI-enabled device must still meet MDR/IVDR requirements on its normal timeline, with the AI Act high-risk layer attaching at the 2028 date. Treat the AI Act as a documented design-input constraint now, because conformity-assessment evidence is cheaper to generate alongside the MDR technical file than to reconstruct afterwards.
What must compliance teams do before August 2, 2026?
Three actions sit on the critical path. First, inventory every AI system in use or pilot across R&D, regulatory affairs, quality and commercial, and classify each against the AI Act's risk tiers. Second, for high-risk systems outside the Annex I deferral, confirm the Article 6(1) Annex III triggers and start the technical documentation and logging baseline before August 2, 2026. Third, for AI-enabled devices and IVDs, align the MDR/IVDR technical file with the AI Act high-risk requirements so the 2028 dual assessment is a continuation, not a restart.
| Date | What becomes applicable | Who is affected |
|---|---|---|
| February 2, 2025 | Prohibited AI practices (Article 5) | All AI providers and deployers in the EU |
| August 2, 2025 | GPAI obligations, notifying authority designation | GPAI model providers, Member State authorities |
| August 2, 2026 | General applicability, transparency duties, remaining Annex III high-risk obligations | All AI system providers and deployers, including life-sciences ops AI |
| December 2, 2026 | Post-Omnibus intimate-imagery prohibition and Article 50(2) watermarking | Providers of relevant content tools |
| August 2, 2028 | Article 6(1) Annex I high-risk for AI-enabled medical devices and IVDs (conditional on Digital Omnibus adoption before August 2, 2026) | MDR/IVDR manufacturers of AI-enabled devices, notified bodies |
Continuous, per-jurisdiction real-time monitoring in Obsidian surfaces register changes the moment they appear, so the August 2, 2026 milestone never arrives unannounced.
Take advantage of this real-time watch
Next steps: confirm which AI systems fall outside the Annex I deferral and must be high-risk compliant by August 2, 2026; brief your notified body and QA leads on the 2028 Annex I timeline and the Digital Omnibus adoption condition; and run a one-page applicability check against MDCG 2025-6.
