Art.111
EU AI Act Guide › Chapter XIII — Final Provisions › Article 111

Article 111 – AI Systems Already Placed on the Market or Put into Service

General SME Relevant ~3 min read · 597 words

Article 111 is the grandfathering provision — one of the most practically important articles for businesses that already have AI systems deployed. It defines the transitional arrangements for AI systems that were already on the market before the Act’s application dates, protecting existing deployments from immediate compliance obligations while setting clear deadlines for achieving compliance.

! High compliance impact for SMEs

WHAT THE ARTICLE IS ABOUT

Transitional protection for existing AI deployments — now updated by the Digital Omnibus

Article 111 establishes the transitional arrangements determining when the AI Act applies to AI systems already in use. The original application dates have been amended by the Digital Omnibus on AI, provisionally agreed on 7 May 2026. Pending formal adoption — expected before 2 August 2026 — the new dates are the operative compliance horizon. The delay does not affect obligations already in force.

WHAT IT SAYS

Updated deadlines following the Digital Omnibus provisional agreement

  • Annex III high-risk AI systems (standalone — employment, education, biometrics, critical infrastructure, law enforcement, justice, migration, essential services): full obligations apply from 2 December 2027, extended from the original 2 August 2026
  • Annex I high-risk AI systems (embedded as safety components in regulated products — medical devices, machinery, etc.): full obligations apply from 2 August 2028, extended from the original 2 August 2027
  • The grace period under Article 111(2) is clarified to cover broader categories of types and models of high-risk AI systems — not each individual unit
  • High-risk AI systems already on the market before the new application dates are subject to the Act only if they undergo significant changes in design or intended purpose from those dates onwards
  • The prohibition on AI practices in Article 5 was not delayed — it applied from 2 February 2025 and remains in full force; the Omnibus adds a new prohibition on AI generating non-consensual intimate imagery and CSAM, applicable from 2 December 2026
  • GPAI obligations under Chapter V were not delayed — they applied from 2 August 2025 and remain unchanged
  • Governance provisions remain in force as originally scheduled
  • High-risk AI systems used by public authorities have until 2 August 2030 to comply
  • The Digital Omnibus is a provisional agreement requiring formal Council and Parliament votes and publication in the Official Journal before 2 August 2026 — formal adoption is expected within weeks

WHO IS AFFECTED

All businesses with AI systems deployed or being developed

  • Providers of Annex III high-risk AI systems — the 16-month delay gives additional time to build compliance infrastructure
  • Providers of Annex I product-embedded high-risk AI — a 12-month extension on top of the original additional year
  • Deployers of high-risk AI systems — the grace period protects existing deployments
  • GPAI model providers — no change to their August 2025 obligations
  • Any operator with systems conducting Article 5 prohibited practices — no delay applies, these must have ceased from February 2025

WHAT IT MEANS FOR SMES

More time — but the clock is running and the goalposts have stopped moving

  • The delay to 2 December 2027 for Annex III systems is real and significant — but every credible source, including the institutions themselves, has signalled this is the only extension you will get; treat it as confirmation of the deadline, not an invitation to pause
  • Use the additional time productively: build your QMS, technical documentation, risk management system, and post-market monitoring framework now — these take longer to implement than most businesses expect
  • The SME relief provisions have been extended to small mid-cap enterprises (fewer than 750 employees, under €150 million turnover) — if you fall in this band, you now benefit from the facilitated procedures and proportionality protections previously limited to SMEs
  • The prohibited practices prohibition is live and unchanged — audit your AI uses against Article 5 immediately if you have not done so
  • The new nudifier and CSAM prohibition under Article 5 applies from 2 December 2026 — if any of your AI systems generate synthetic intimate imagery, you have until that date to ensure compliance

Related Articles

← Back to the Guide
← Previous Art. 110 — Amendment to Directive (EU) 2020/1828 Next → Art. 112 — Evaluation and Review