That depends on one input this page does not decide: the classification. Pick it below — decided by the engine or stated by you — and every date renders as one sentence carrying the provision that sets it. Until you pick, the whole map is visible.
The Regulation as a whole applies from 2 August 2026 under Article 113, second paragraph.
Applies regardless of the classification input.
The Article 5 prohibitions have applied since 2 February 2025 under Article 113, third paragraph, point (a) — with the exception written into the same point: Article 5(1), first subparagraph, points (ba) and (bb), and Article 5(1a) and (1b) apply from 2 December 2026.
Applies regardless of the classification input.
The Article 4 AI literacy duty has applied since 2 February 2025 — it sits in Chapter I, which Article 113, third paragraph, point (a) put on that date.
Applies regardless of the classification input.
The Article 50 transparency duties apply from 2 August 2026: Chapter IV sits in no exception, so it takes the general date in Article 113, second paragraph — save that Article 111(4) gives providers of systems generating synthetic content that were placed on the market before 2 August 2026 until 2 December 2026 to comply with Article 50(2).
Applies regardless of the classification input.
The high-risk regime — Chapter III, Sections 1, 2 and 3, with the exception of Article 6(5) — applies from 2 December 2027 to systems classified as high-risk pursuant to Article 6(2) and Annex III (Article 113, third paragraph, point (c), as amended by Regulation (EU) 2026/1744).
For a system classified as Annex III high-risk — the classification is the input, not the finding.
Article 43 (conformity assessment) and Article 49 (registration) sit in Chapter III, Section 5, which point (c) does not defer — so they keep the general date in Article 113, second paragraph: 2 August 2026.
For a system classified as Annex III high-risk — the classification is the input, not the finding.
How Section 5 is meant to operate while the Section 1 classification rules are not yet in application is a question the amending regulation does not answer. We state it as open rather than resolve it.
For a system classified as Annex III high-risk — the classification is the input, not the finding.
The high-risk regime — Chapter III, Sections 1, 2 and 3, with the exception of Article 6(5) — applies from 2 August 2028 to systems classified as high-risk pursuant to Article 6(1) and Annex I (Article 113, third paragraph, point (c), as amended by Regulation (EU) 2026/1744).
For a system classified as Annex I high-risk — the classification is the input, not the finding.
The rest of the Annex I map — the sectoral conformity route and the notified-body dates — is in the deadlines table, row by row.
For a system classified as Annex I high-risk — the classification is the input, not the finding.
No high-risk date attaches. What binds you are the dates every operator carries — the rows above. A classification is a dated answer: if the system changes, the question is asked again.
For a system classified as not high-risk — the classification is the input, not the finding.
The obligations for providers of general-purpose AI models in Chapter V have applied since 2 August 2025 under Article 113, third paragraph, point (b) — a point whose own text excepts Article 101.
For a system classified as a general-purpose AI model — the classification is the input, not the finding.
A general-purpose AI model placed on the market before 2 August 2025 must be brought into compliance by 2 August 2027 under Article 111(3), which Regulation (EU) 2026/1744 did not touch.
For a system classified as a general-purpose AI model — the classification is the input, not the finding.
An operator of a high-risk AI system placed on the market or put into service before the date of application of Chapter III referred to in Article 113 falls under the Regulation only if, as from that date, the system is subject to significant changes in its design — Article 111(2), as replaced by Regulation (EU) 2026/1744. Whether a change is significant is a judgment, and this page does not make it.
A condition, not a question: whether it holds for your system is not this page's call.
The replaced paragraph refers in the singular to the date of application of Chapter III, and Article 113 now carries two such dates — 2 December 2027 and 2 August 2028. Which one governs a given legacy system is not resolved by the text, and we say so rather than pick one.
A condition, not a question: whether it holds for your system is not this page's call.
In any case, providers and deployers of high-risk AI systems intended to be used by public authorities must comply with the Regulation's requirements and obligations by 2 August 2030 — the second sentence of Article 111(2), which kept that date through the amendment.
A condition, not a question: whether it holds for your system is not this page's call.
A provider of an AI system that generates synthetic audio, image, video or text content and that was placed on the market before 2 August 2026 must comply with Article 50(2) by 2 December 2026 under Article 111(4), added by Regulation (EU) 2026/1744. Whether a system generates such content is a fact about the system — an input this page does not take.
A condition, not a question: whether it holds for your system is not this page's call.
The Annex X large-scale IT systems carry their own transitional rule in Article 111(1); it is in the deadlines table.
A condition, not a question: whether it holds for your system is not this page's call.
No dates render here, because every date on this page attaches to a classification, and a guessed classification produces confident, wrong deadlines. The classifier decides it in your browser, and each answer cites the article it rests on.
This tool keeps no record of what you pick: its script stores nothing and sends nothing, and the build reads the delivered page on every run and stops if that changes. The full table, including the rows this page does not render: every EU AI Act date, with the article behind it.
Complipath is EU AI Act compliance software for AI-heavy software companies without a compliance team — an AI system register, deterministic risk classification, the obligations that follow, and the evidence behind every decision.
Complipath is built by Yobel Tzegai in Gothenburg, Sweden.
Complipath provides legal information, not legal advice. Every guide cites its source on EUR-Lex — Regulation (EU) 2024/1689, and Regulation (EU) 2026/1744 where that has amended it; where the law is still settling, the guide says so.
We measure page views with Vercel Web Analytics. It uses no third-party cookies. Visitors are identified by a hash derived from the incoming request, which is discarded after 24 hours, and no identifier is stored that could follow a visitor to another site. What is collected: the time of the visit, the URL, the referring page, filtered query parameters, city-level location, operating system, browser and device type.