AI Act
The AI Act applies in stages; after the Digital Omnibus, high-risk rules apply from 2 December 2027 (Annex III) and 2 August 2028 (AI in regulated products).
- Reference
- Regulation (EU) 2024/1689, amended by Regulation (EU) 2026/1744
- Status
- Applies in stages
- Sources last checked
Content reviewed by DI(FH) Herbert Schrank · sources last checked
Key dates — get ready
As ofNext date: (in 29 days) — New prohibitions (AI-generated non-consensual intimate content, child sexual abuse material) and marking of AI-generated content for systems placed on the market before 2 August 2026
- 1Entry into force — obligations apply in stagesPassed
- 2Prohibited practices and AI literacyPassed
- 3General-purpose AI rules and governancePassed
- 4General application incl. transparency duties (marking of generated content for certain systems: 2 Dec 2026)Passed
- 5New prohibitions (AI-generated non-consensual intimate content, child sexual abuse material) and marking of AI-generated content for systems placed on the market before 2 August 2026Upcoming
- 6High-risk systems in Annex III areas (postponed by (EU) 2026/1744)Upcoming
- 7High-risk AI in regulated products (Annex I)Upcoming
Get ready in time — check which obligations apply to your products.
Our sister brand for AI governance and the EU AI Act — same owner (website in German).
What it is
The AI Act regulates AI systems by risk. Prohibited practices and AI-literacy provisions apply since February 2025 (the AI-literacy provision was softened by Regulation (EU) 2026/1744), rules for general-purpose AI since August 2025, and transparency duties since August 2026. Regulation (EU) 2026/1744 (in force since 27 July 2026) postponed the high-risk obligations and moved AI requirements for machinery into the Machinery Regulation.
Who is affected
Providers, importers, distributors and deployers of AI systems — including AI functions embedded in devices, machines and software.
What you have to do
- Inventory AI functions in your products and processes and classify them by risk.
- Take measures to support the AI literacy of staff dealing with AI systems (Art. 4 as amended by Regulation (EU) 2026/1744).
- Providers: mark AI-generated output in machine-readable form and design systems so that people are informed they are interacting with AI; deployers: disclose deepfakes and certain AI-generated text.
- Prepare high-risk documentation in time for December 2027 / August 2028; for the quality management system (Art. 17), the harmonised standard EN 18286:2026 can be used.
Fact check
Not correct “The AI Act has been postponed as a whole.”
Mainly the high-risk obligations were postponed (plus, for systems placed on the market before 2 August 2026, the marking of generated content until 2 December 2026). Prohibitions, AI-literacy measures, general-purpose AI rules and the other transparency duties already apply.
Sources
- Regulation (EU) 2024/1689 (AI Act) — Official Journal text (opens external website) ↗
- Regulation (EU) 2026/1744 (Digital Omnibus on AI) — Official Journal text (opens external website) ↗
- AI Act — timeline incl. omnibus changes (opens external website) ↗
- Implementing Decision (EU) 2026/2274 — harmonised standards for high-risk AI systems (opens external website) ↗
- First harmonised standard supporting the AI Act published in the Official Journal (EN 18286:2026) (opens external website) ↗
- Guidelines on the transparency obligations under Art. 50 AI Act (C(2026) 5054) (opens external website) ↗
- Commission publishes Code of Practice on marking and labelling of AI-generated content (opens external website) ↗
General information, not legal advice. Authentic are only the texts published in the Official Journal of the EU and the Austrian Federal Law Gazette.
Not sure where your product stands?
Send us the product and its documents. We check the technical file against the rules on this radar and tell you what is missing — and source an EU-ready alternative if needed.