欧盟

European Union comprehensive The EU AI Act (Regulation 2024/1689) entered into force on 1 August 2024. It creates harmonised rules for placing AI on the EU market; applies to EU and third-country providers and deployers placing AI systems on the EU market; centres on a risk-based approach; prohibits certain AI practices and imposes specific requirements on high-risk systems; and creates harmonised transparency rules. 2 Feb 2025 — prohibitions on unacceptable-risk practices and the AI literacy obligation applied. 2 Aug 2025 — obligations for general-purpose AI models, governance provisions, penalties architecture and member-state notification/surveillance authority designation applied. 2 Aug 2026 — Article 50 transparency obligations apply, with penalties of up to €15m or 3% of worldwide annual turnover. Enforcement is split: national market surveillance authorities enforce most of the Act; the AI Office supervises general-purpose AI models and systems built on them by the same provider. 10 Jul 2025 — General-Purpose AI Code of Practice published The final GPAI Code of Practice — three chapters covering transparency, copyright, and safety & security — was published after input from 1,400+ stakeholders, and formally approved by the Commission and AI Board on 1 August 2025. Formally voluntary, it is the presumptive route to demonstrating compliance with Articles 53 and 55; most major model providers signed. European Commission — General-Purpose AI Code of Practice <https://digital-strategy.ec.europa.eu/en/policies/contents-code-gpai> 2 Aug 2025 — GPAI obligations apply Providers of general-purpose AI models placed on the market from this date must comply immediately. Commission enforcement powers (information requests, model access, recalls) become exercisable from 2 August 2026. Models placed on the market before 2 August 2025 have until 2 August 2027 to comply. Regulation (EU) 2024/1689 (AI Act) — Official Journal <https://eur-lex.europa.eu/eli/reg/2024/1689/oj>

24 Jul 2026 — Digital Omnibus published in the Official Journal as Regulation (EU) 2026/1744 Publication in the OJ on 24 July 2026 fixed the amendments in law — the first formal amendment of the AI Act since its adoption in June 2024. Regulation (EU) 2026/1744 (Digital Omnibus on AI) — Official Journal <https://eur-lex.europa.eu/eli/reg/2026/1744/oj> 27 Jul 2026 — Digital Omnibus on AI enters into force Three days after publication the amending Regulation took effect. Annex III stand-alone high-risk obligations move from 2 August 2026 to 2 December 2027; Annex I embedded high-risk systems move from 2 August 2027 to 2 August 2028 — fixed dates rather than the standards-contingent trigger the Commission originally proposed. The Omnibus also adds new prohibitions on placing on the market or using AI systems designed to generate child sexual abuse material or non-consensual sexual or intimate imagery — applying from 2 December 2026, not on entry into force; rewrites the Article 4 AI literacy duty from an obligation to ensure a sufficient level into an obligation to take measures to support it, still binding on every deployer and subject to national supervision from 3 August 2026; and broadens the AI Office's supervisory role over general-purpose AI models, aligning parts of it with the Digital Services Act model. Regulation (EU) 2026/1744 (Digital Omnibus on AI) — Official Journal <https://eur-lex.europa.eu/eli/reg/2026/1744/oj>

17 Sep 2026 — Commission proposes the EU KIDS Act — AI companions off by default for minors The proposed Keeping Internet Digital Spaces Accountable and Trustworthy Act covers social networks, video platforms, games, app stores, operating systems and, explicitly, AI companions and conversational chatbots. For minors, chatbots and companions would be switched off by default, must not simulate relationships in ways likely to create emotional dependency, and must not carry memory between sessions by default; providers would have to evaluate risks to minors' health, safety and fundamental rights. A proposal only — it now goes to Parliament and Council — but it would be the first EU instrument to regulate companion AI by design. European Commission — Proposal for an EU KIDS Act (Keeping Internet Digital Spaces Accountable and Trustworthy), 17 Sep 2026 <https://digital-strategy.ec.europa.eu/en/library/proposal-eu-kids-act-eu-keeping-internet-digital-spaces-accountable-and-trustworthy>

EU KIDS Act (Commission proposal, 17 Sep 2026) [draft]

8 Jul 2026 — EDPB opens consultation on web scraping for generative AI The European Data Protection Board published draft Guidelines 03/2026 on web scraping in the context of generative AI, clarifying how the GDPR applies where personal data is collected from the internet at scale to develop or train models. The draft confirms the GDPR applies to mixed personal and non-personal datasets, works through controller/processor roles, purpose limitation and data minimisation, and treats legitimate interests as the usual lawful basis — subject to a three-part test weighing the scale of collection, the source, whether anti-scraping measures were in place, and whether narrower collection, synthetic or pseudonymised data would do. Transparency is not waived by scale: the Article 14(5)(b) disproportionate-effort exemption is not a blanket carve-out, and controllers must still publish detail on sources, purposes and how to exercise rights. There is no general exemption for special category data collected incidentally. Consultation runs to 30 October 2026. This is data protection law, not the AI Act — but it is where training-data practice is actually being set in the EU. EDPB — Guidelines 03/2026 on web scraping in the context of generative AI (public consultation) <https://www.edpb.europa.eu/public-consultations/guidelines-032026-on-web-scraping-in-the-context-of-generative-ai_en>