Every AI Policy Ever · open in the interactive register

Published (non-binding, current)Non-binding · Guidelines / framework

Commission Guidelines on the definition of an artificial intelligence system established by Regulation (EU) 2024/1689 (AI Act)

Record EU-07

Issued by
European Commission
Jurisdiction
European Union
Level · region
Supranational · Europe
Published / adopted
2025-02-06
Last updated
2026-04-08
Language
English

Summary

Non-binding Commission guidelines that explain how to apply the AI Act's legal definition of an 'AI system', to help providers and other relevant persons determine whether a software system is an AI system and so falls within the Act's rules, which sort AI systems into risk categories: prohibited practices, high-risk systems and systems subject to transparency obligations. They were issued as the definition, AI literacy and the prohibitions began to apply on 2 February 2025, and are to be updated in light of practical experience, new questions and use cases.

Safety topics

risk-based classificationprohibited practices

Status notes

Published on 6 February 2025, in addition to the guidelines on prohibited AI practices (EU-06), to facilitate the application of the AI Act's first rules, which have applied since 2 February 2025. The Commission approved the draft guidelines by a Communication; they are available in the 24 official EU languages. The Commission says they will be updated as necessary, in light of practical experience, new questions and use cases; the page shows a last update of 8 April 2026 without saying what changed.

Sources

Open the official source https://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-ai-system-definition-facilitate-first-ai-acts-rules-application

Local copy

Archived. A source changed since the previous snapshot; both versions are kept. Last attempt 2026-10-17 18:20:39 UTC

Official source (third-party copy, not published) archive/EU-07/2026-10-11/official.html
Retrieved 2026-10-11 13:58:18 UTC
SHA-256 b9e8cf5f88adfb3ae501728a1c9d732f730f5d5037820305d13eb25e49572795
Linked document (PDF) (third-party copy, not published) archive/EU-07/2026-10-17/document-1.pdf
Retrieved 2026-10-17 16:46:44 UTC
SHA-256 7eff30775b3ba68abec42a7d14a66433657e1400f812af1c84faccc11ea69d46

Verification

A
The instrument's own official text, or an official register, was stored and confirms the title, date and status. checked 2026-10-17 · official-url-fetched
Stored and read in this run: the Commission's library page for the guidelines (snapshot of 11 Oct 2026, unchanged on 17 Oct), which gives the title, publication on 6 February 2025, the non-binding nature, the 24 language versions and a last update of 8 April 2026, and its printable PDF. The guidelines themselves and the approving Communication are not stored, because the site's robots.txt does not allow automated downloads of them; the summary rests on the library page, and official_identifier (the Communication's C(2025) number) stays null. The summary was below the 250-character minimum and has been extended from the page; 'prohibited practices' was added as a second topic because the page presents the guidelines as support for the first rules, which include the prohibitions. Version 1's remark that the C(2025) number was not verified moved here from the notes.
Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-17: 7 field(s) supported.

History

  1. 2026-10-17notes: Published 6 Feb 2025, together with the prohibited-practices guidelines. The Com → Published on 6 February 2025, in addition to the guidelines on prohibited AI pra — research remark on the C(2025) number moved to verification.note; 'together with' replaced by the page's 'in addition to'; language versions added2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
  2. 2026-10-17safety_topics: ["risk-based classification"] → ["risk-based classification", "prohibited practices"] — a second topic was needed; 'prohibited practices' added: the page presents the guidelines as support for the first rules, which include the prohibitions, and they were issued with the prohibited-practices guidelines2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
  3. 2026-10-17summary: Non-binding guidance that explains the elements of the Art. 3(1) definition of a → Non-binding Commission guidelines that explain how to apply the AI Act's legal d — version 1's summary was below 250 characters; extended from the stored page (purpose, risk categories, the first rules of 2 February 2025, updates)2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
  4. 2026-10-14language: English (later translated into other EU languages) → English — The field held a remark after the language name; it now gives the language names only (AGENTS.md 5.6).manual-2026-10-14 · Claude Code (claude-opus-5-5)

Related records

EU AI Act European Union · Partially in force

Guidelines on prohibited AI practices European Union · Published (non-binding, current)

Citation

European Commission. Commission Guidelines on the definition of an artificial intelligence system established by Regulation (EU) 2024/1689 (AI Act). 2025-02-06. https://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-ai-system-definition-facilitate-first-ai-acts-rules-application. Status: Published (non-binding, current); verified 2026-10-17 (level A). Local copy archive/EU-07/2026-10-11/official.html retrieved 2026-10-11 13:58:18 UTC, SHA-256 b9e8cf5f88adfb3ae501728a1c9d732f730f5d5037820305d13eb25e49572795. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record EU-07.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.