EU AI Act enforcement log
No. As at 28 October 2026, no fine under the EU AI Act appears on any official EU page we checked, and no formal proceeding (the procedural step that can lead to a Commission fine) has been announced. The first documented enforcement steps are requests for information that the AI Office sent to providers of the most advanced general-purpose AI models, which the Commission confirmed in October 2026 and described as its "first enforcement actions". The Commission has also said it received serious incident reports. We have not found any AI Act enforcement action by a national authority on an official page.
Every action and every finding of absence below links to an official source: the European Commission, the AI Office, the Official Journal or a national authority. Reports we could not confirm on an official page are left out. This is an information service, not legal advice; see how we check.
Documented actions
| Date | Who | What happened | What it is not | Source |
|---|---|---|---|---|
| 2026-10-14 | European CommissionEU | Request for informationIn a speech in Dublin, Executive Vice-President Virkkunen said the Commission had "already taken our first enforcement actions under the AI Act": the AI Office "sent Requests for Information addressing issues such as security, external evaluation and monitoring how models are used".Scope: Providers of general-purpose AI models | "Enforcement actions" is the Commission's wording for the requests for information. No formal proceeding, decision or fine is described, and no recipient is named. | SPEECH/26/2150 |
| 2026-10-12 | AI OfficeEU | Request for informationA Commission news article said the AI Office "has already requested and received serious incident reports and sent requests for information on security practices", and that the Commission "will send requests for information with the combined questions of the panel and the AI Office".Scope: Providers of general-purpose AI models | The article says those requests are meant to "further investigate these incidents", but it describes no formal proceeding under Regulation (EU) 2026/1755, no decision and no fine. No company is named. | Special meeting of the AI Scientific Panel |
| 2026-10-09 | AI OfficeEU | Request for informationOpening a meeting of the AI Scientific Panel on 9 October (text published 13 October), Executive Vice-President Virkkunen said the AI Office had sent its first requests for information to providers of the most advanced models, on security, external evaluation and post-market monitoring, and was receiving replies.Scope: Providers of the most advanced general-purpose AI models | No send date, recipient count or recipient names are given for the requests. No formal proceeding, decision or fine. | SPEECH/26/2148 |
| 2026-10-09 | AI OfficeEU | Serious incident reports receivedThe same speech said the Commission had "already received serious incident reports" and was analysing them, and that, as a next step, "the AI Office will issue a request for information about the recent incidents".Scope: Providers of general-purpose AI models | Receiving a report is not a finding of an infringement. The speech announces the incident request as a next step; it does not say it has been sent. | SPEECH/26/2148 |
| 2026-10-09 | European CommissionEU | OtherThe Commission announced a special meeting of the AI Scientific Panel, held on 9 October, on recent loss-of-control incidents; it said the panel, together with the AI Office, "has worked on a set of questions for the companies that developed the models involved".Scope: Providers of general-purpose AI models | The page describes questions being prepared; it states no request sent, no formal proceeding and no fine. The "investigating" it mentions is the panel's own work. | Commission holds special meeting of Scientific panel on frontier AI safety and risks |
What official pages do not show
- As at 28 October 2026: the Commission's AI Act enforcement page (last updated 20 October 2026) describes enforcement powers and the timeline, and reports no fine, decision or opened proceeding. (Enforcement of the AI Act)
- As at 28 October 2026: the Commission's news feed and press corner contain no AI Act fine, decision or opened proceeding; the only enforcement items in the news feed after 12 October 2026 concern other EU laws. (Commission press corner)
- As at 28 October 2026: the Commission's notified-bodies register, filtered to the AI Act, lists 0 active notified bodies, so no conformity assessment body has yet been notified under the AI Act. (Notified bodies register, filtered to the AI Act)
When each enforcement power applies
- 2025-08-02The penalties chapter applies (Articles 99 and 100), except the Commission's fines for general-purpose AI model providers. The general-purpose AI model obligations, the notified-body rules and the governance chapter apply from the same date. (Art. 113, point (b))
- 2025-08-02By this date Member States had to notify the Commission of their national penalty rules (our reading: Article 99(2) says "by the date of entry into application", and the penalties chapter applies from 2 August 2025). (Art. 99(1)–(2) with Art. 113(b))
- 2025-08-02Providers of general-purpose AI models with systemic risk must report serious incidents to the AI Office without undue delay. (Art. 55(1)(c))
- 2026-08-02The Commission's enforcement powers over general-purpose AI model providers apply: requests for documents and information, evaluations, measures, and fines of up to 3% of worldwide annual turnover or EUR 15 million, whichever is higher. Before a fine, the Commission must communicate preliminary findings and give the provider an opportunity to be heard. (Arts. 91–94 and 101(1)–(2); Art. 113, point (b))
- 2026-08-02The rest of the Regulation applies, including the market-surveillance chapter (Chapter IX). The Commission's enforcement page says the enforcement powers of the AI Office and of national authorities apply from this date. (Art. 113, first paragraph)
- 2026-08-10Commission Implementing Regulation (EU) 2026/1755 enters into force (20th day after its publication in the Official Journal on 21 July 2026), setting detailed arrangements for Commission evaluations of general-purpose AI models and for proceedings that can lead to fines. (Reg. (EU) 2026/1755, Art. 15; adopted under AI Act Art. 92(6) and 101(6))
- 2026-12-02The two new Article 5 prohibitions added by the Digital Omnibus apply. (Art. 113(a) as amended)
- 2027-12-02High-risk requirements apply to Annex III systems, so they can only be enforced from this date. (Art. 113(c)(i) as amended)
- 2028-08-02High-risk requirements apply to AI systems embedded in products covered by Annex I. (Art. 113(c)(ii) as amended)
Questions
Has anyone been fined under the EU AI Act?
Not as far as official sources show. As at 28 October 2026, no AI Act fine, decision or formally opened proceeding appears on the European Commission's AI Act enforcement page, the AI Office page, the Commission's news feed or its press corner, and we found no national authority action on an official page.
What enforcement action has the AI Office taken?
In October 2026 the Commission confirmed that the AI Office had sent its first requests for information to providers of the most advanced general-purpose AI models, on security, external evaluation and post-market monitoring, and that it had received serious incident reports. It called these its first enforcement actions under the AI Act. No formal proceeding, decision or fine has been announced, and no recipient has been named officially.
When can fines under the EU AI Act be imposed?
The penalties chapter (Chapter XII) has applied since 2 August 2025, and the Commission's enforcement page says the enforcement powers of the AI Office and of national authorities apply from 2 August 2026. From then, national authorities can in principle sanction infringements of obligations that already apply, such as the Article 5 prohibitions, but only under the penalty rules each Member State has adopted and through the authorities it has designated. The Commission's own power to fine providers of general-purpose AI models (Article 101) applies from 2 August 2026, and before fining it must communicate preliminary findings and give the provider an opportunity to be heard (Article 101(2)). The main high-risk obligations apply only from 2 December 2027 (Annex III) or 2 August 2028 (Annex I).
For who supervises what, see who enforces the EU AI Act; for the fine levels, see how much AI Act fines can be. To check which obligations apply to your company, answer a few questions.
This is an information service, not legal advice.