European Union

• 2026-10-10:Aggregators reported that the Commission has sought information from more than 30 AI companies. This isreported, not confirmed: no official release was located and it is not treated as a new action.
• 2026-10-06:The Commission refreshed its AI Act enforcement page(Article 50, GPAI enforceability, fines, complaint and whistleblower tools).
• Late August 2026:The AI Office reportedly sent its first requests for information to GPAI providers (about 2026-08-29; no official press URL retrieved).

What to watch:2026-12-02 (new prohibitions apply and the Article 50(2) grace ends), the first harmonised standards, and any official statement on the reported information requests.

The core instrument is the AI Act, Regulation (EU) 2024/1689, as amended by the Digital Omnibus on AI, Regulation (EU) 2026/1744 (signed 2026-07-08, published in the Official Journal 2026-07-24, in force 2026-07-27). Related instruments are:

• Omnibus deferrals.Obligations for Annex III high-risk systems are deferred to 2027-12-02 and for Annex I (product-safety) systems to 2028-08-02.
• New prohibitions.Two new statutory points in Article 5, points (ba) and (bb), prohibit non-consensual sexual or intimate content and child sexual abuse material, including “nudification” tools. They apply from 2026-12-02. (An overview page groups them as one “ninth category”; they are two statutory points.)
• Code of Practice on marking and labelling AI-generated content.Voluntary; published around 2026-06-10.
• EDPB–EDPS Joint Opinion 1/2026on the Omnibus proposal (2026-01-20); advisory.
• EU Action Plan on Cybersecurity and AI(July 2026).
• Council of Europe Framework Convention(CETS 225).The EU ratified it on 2026-05-15; it is not yet in force (see the Council of Europe entry).

The AI Act is directly applicable and has extraterritorial reach. It applies to providers and deployers of AI systems, uses a risk-tier structure, and regulates general-purpose AI models and systems built on them, with oversight centralised in the AI Office. The Omnibus adds simplifications for small and medium-sized enterprises and small mid-cap companies and an EU-level regulatory sandbox.

Member-state implementation is tracked separately: see the Germany, France, the Netherlands and Spain entries.

Administrative finesare tiered: up to EUR 35 million or 7% of worldwide turnover for prohibited practices; up to EUR 15 million or 3% for other breaches, including GPAI obligations; and up to EUR 7.5 million or 1% for supplying incorrect information in response to information requests. The AI Officeand national authorities have held enforcement powers since 2026-08-02, and the Commission provides complaint and whistleblower tools.

• Article 50 transparency(chatbot disclosure, deepfake labelling, machine-readable marking) applies from 2026-08-02. Article 50(2) has a legacy-system grace period to 2026-12-02 under Article 111(4) for systems placed on the market earlier; it is not a blanket delay for generative AI.
• Standards.No harmonised standard had been cited in the Official Journal as of about 2026-08-24, so the presumption-of-conformity route is not yet available.

Caveat:the authentic Official Journal text of Regulation 2026/1744 was not accessed (EUR-Lex was degraded during verification), so article-level statements rest on official Commission pages and should be re-verified against the OJ. Research assistance, not legal advice.