Every AI Policy Ever · open in the interactive register
In forceBinding · Regulation / implementing rule
Commission Implementing Regulation (EU) 2025/454 of 7 March 2025 laying down the rules for the application of Regulation (EU) 2024/1689 of the European Parliament and of the Council as regards the establishment of a scientific panel of independent experts in the field of artificial intelligence
Record EU-08 · Implementing Regulation (EU) 2025/454
- Issued by
- European Commission
- Jurisdiction
- European Union
- Level · region
- Supranational · Europe
- Published / adopted
- 2025-03-07
- In force
- 2025-03-30
- Language
- All 24 official EU languages
- Official identifier
- Implementing Regulation (EU) 2025/454
Summary
Establishes, under Article 68(1) of the AI Act, a scientific panel of independent experts in AI to support enforcement and advise the AI Office. The panel has at most 60 experts, appointed for two-year terms after a public call, with one to three nationals of each EU and EEA-EFTA state where qualified applicants exist; they must be independent of providers of AI systems and general-purpose AI models, declare their interests and work transparently. Market surveillance authorities can request the panel's support, and the Regulation sets the procedures for the panel to issue qualified alerts to the AI Office and to request information from providers of general-purpose AI models.
Safety topics
general-purpose / frontier AIevaluation & testing
Status notes
Adopted on 7 March 2025 (Commission document C/2025/1377) and published in the Official Journal (OJ L, 2025/454) on 10 March 2025; it entered into force on 30 March 2025, the twentieth day after publication. The AI Office and the Joint Research Centre jointly provide the panel's secretariat. On 1 June 2026 the Commission appointed the Scientific Panel, with 60 independent experts, and the Advisory Forum, for two-year terms; the Panel focuses on general-purpose AI models and systems, systemic risks, model classification, evaluation methodologies and cross-border market surveillance. On 9 October 2026 the Commission held a special meeting of the Scientific Panel on frontier AI safety and risks, at which the panel was to present recommendations on frontier AI safety and security risks; the panel had been investigating recent loss-of-control incidents and, with the AI Office, had worked on a set of questions for the companies that developed the models involved.
Local copy
Archived. Saved with a headless browser because the plain download looked incomplete. A source changed since the previous snapshot; both versions are kept. Last attempt 2026-10-17 18:20:41 UTC
Official source (third-party copy, not published) archive/EU-08/2026-10-17/official.html
Retrieved 2026-10-17 18:20:45 UTC
SHA-256 dc6faaaa95acf0338a992c40a47858e626dc6218c3cfe1d2fddd94430b93e5c7
Warning: The site returned a JavaScript bot-check page (web application firewall) instead of the document
Official source, rendered in a browser (third-party copy, not published) archive/EU-08/2026-10-17/official-rendered.html
Retrieved 2026-10-17 18:19:24 UTC
SHA-256 84442ec8012d0f8bc8fb9eaf82d8603b2be52e005ec1b4e233aaecb25194a14a
Official source, printed to PDF by a browser (third-party copy, not published) archive/EU-08/2026-10-17/official-print-2.pdf
Retrieved 2026-10-17 18:21:39 UTC
SHA-256 3922e3bd2d6a0a40b9caf233f656c64581e621ed2a8da7e6213f9fe2413ad720
Secondary source (third-party copy, not published) archive/EU-08/2026-10-11/secondary-2.html
Retrieved 2026-10-11 14:07:36 UTC
SHA-256 92f12b0b8692e622389e2f5489bfa63145b52ba51d2795b55e629f6fbd24127a
Verification
AThe 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 EUR-Lex page with the Official Journal text (snapshot of 11 Oct 2026, unchanged on 17 Oct), which gives the full title, the document number C/2025/1377, publication in OJ L 2025/454 of 10 March 2025, adoption at Brussels on 7 March 2025, Article 20 (entry into force on the twentieth day after publication) and the legal status 'In force' with no notice of changes; the Commission's press release of 1 June 2026 (alt-official) on the appointment of the Scientific Panel and the Advisory Forum; and the Commission's press release on the special meeting of the Scientific Panel (17 Oct 2026 copy), published 9 October 2026, with a related item saying the meeting was convened on 9 October 2026. That press release confirms version 1's date of 9 Oct 2026 and its account of the loss-of-control incidents and the questions for developers; the first pass of this run had wrongly redated the meeting to 5 October 2026, the date the news list of the Commission's AI Act page gives the item. Version 1's date_last_updated of 1 June 2026 was the date of the appointments, not an amendment of the Regulation; EUR-Lex shows no amendment, so the field is now null. 'Catastrophic risk' and 'safety institute' were dropped: the Regulation does not mention catastrophic risks, and the panel is an advisory expert body for enforcement, not an AI safety institute.
Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-17: 11 field(s) supported.
History
- 2026-10-17notes: Adopted 7 March 2025. Published in OJ L on 10 March 2025. In force 30 March 2025 → Adopted on 7 March 2025 (Commission document C/2025/1377) and published in the O — document number, OJ reference and secretariat added; the special meeting of 9 October 2026, the loss-of-control investigation and the questions for developers are now quoted from the stored press release2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
- 2026-10-17safety_topics: ["general-purpose / frontier AI", "evaluation & testing", "catastrophic risk", " → ["general-purpose / frontier AI", "evaluation & testing"] — 'catastrophic risk' dropped (the Regulation does not deal with it) and 'safety institute' dropped (the panel is an advisory expert body for enforcement, not an AI safety institute); 'general-purpose / frontier AI' (Arts 15 to 18) and 'evaluation & testing' (Art. 7(4), evaluations under Art. 92(2) of the AI Act) stay2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
- 2026-10-17summary: Implements AI Act Art. 68 by setting up a scientific panel of independent AI exp → Establishes, under Article 68(1) of the AI Act, a scientific panel of independen — rewritten from recital 1 and Articles 3, 4, 10, 12, 14, 15 and 182026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
- 2026-10-17date_last_updated: 2026-06-01 → – — set to null: version 1's 1 June 2026 is the date the panel was appointed, not an amendment or new version; EUR-Lex shows the Regulation 'In force' without the 'This act has been changed' notice it shows for amended acts2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
- 2026-10-17title: Commission Implementing Regulation (EU) 2025/454 of 7 March 2025 laying down the → Commission Implementing Regulation (EU) 2025/454 of 7 March 2025 laying down the — the verbatim title from the Official Journal text, which names the AI Act 'Regulation (EU) 2024/1689 of the European Parliament and of the Council'2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
Citation
European Commission. Commission Implementing Regulation (EU) 2025/454 of 7 March 2025 laying down the rules for the application of Regulation (EU) 2024/1689 of the European Parliament and of the Council as regards the establishment of a scientific panel of independent experts in the field of artificial intelligence [Implementing Regulation (EU) 2025/454]. 2025-03-07. https://eur-lex.europa.eu/eli/reg_impl/2025/454/oj/eng. Status: In force; verified 2026-10-17 (level A). Local copy archive/EU-08/2026-10-17/official.html retrieved 2026-10-17 18:20:45 UTC, SHA-256 dc6faaaa95acf0338a992c40a47858e626dc6218c3cfe1d2fddd94430b93e5c7. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record EU-08.
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.