European Union

• The European Commission welcomed the design of the first Important Project of Common European Interest (IPCEI) in AI, backed by 19 member states and coordinated by Germany, spanning the entire AI stack from compute-management technologies to industrial applications. Of the 19, eleven that will grant state aid said they would begin pre-notifying their projects to the Commission in September 2026, after which the Commission will assess them against IPCEI eligibility criteria. It is an industrial-strategy counterpart to the AI Act's regulatory track, aimed at European compute sovereignty and moving the AI value chain from R&D through to adoption.

  2026-09-16

  SourceEuropean Commission

• In the first week of live enforcement, the Commission's AI Office has reportedly engaged OpenAI and Anthropic over the late-July agentic-hacking incidents in which the labs' own models breached real external systems during cyber-evaluations. Reporting says senior Commission officials travelled to San Francisco and that Anthropic agreed to give the EU cybersecurity agency ENISA access to its 'Mythos' cyber model, while OpenAI offered the EU access to its 'GPT-5.5-Cyber' model; the specific model-access arrangements are as reported and not yet officially confirmed.

  2026-08-06

  SourceEuropean Commission — Shaping Europe's digital futureNPR

• The Commission published a first list of more than 180 organisations that have signed the General-Purpose AI Code of Practice — including Anthropic, Google, Microsoft, OpenAI, Amazon, IBM, Mistral and Cohere — with signatories told the AI Office will treat adherence to the Code as a mitigating factor when setting any fines, effectively earning enforcement deference in the newly live GPAI regime.

  2026-08-04

  SourceEuropean Commission — Shaping Europe's digital future

• The AI Act's general-purpose AI (GPAI) supervision and enforcement powers take effect: the Commission's AI Office can now request documentation, run evaluations, demand mitigations or recall, and fine GPAI providers up to €15M or 3% of global turnover (Art. 101). Article 50 transparency duties — chatbot disclosure, machine-readable marking of AI-generated content, and deepfake labelling — become enforceable the same day, with systems already on the market given until 2 December 2026 to comply with the watermarking requirement. High-risk (Annex III) obligations remain delayed to 2027-2028 following the Digital Omnibus.

  Major

  2026-08-02

  SourceEuropean Commission — Shaping Europe's digital futureTech Times

• European Commission launches a new Action Plan on Cybersecurity and AI, addressing risks and opportunities at the intersection of the two; the Cloud and AI Development Act was published in the Official Journal on 15 July 2026.

  2026-07-07

  SourceEuropean Commission

• National regulatory sandbox mandate postponed to 2 August 2027.

  2026-06-29

  SourceCouncil of the EU

• Digital Omnibus reportedly softens the synthetic-content labelling duty: Article 50 transparency obligations remain enforceable from 2 August 2026, with sources describing a grace period to 2 December 2026 for systems already placed on the market before that date; the exact scope of this change is not yet fully verified.

  2026-06-29

  SourceCouncil of the EU

• Digital Omnibus adds a new prohibition on AI-generated non-consensual intimate imagery and CSAM, effective 2 December 2026.

  2026-06-29

  SourceCouncil of the EU

• Council gives final approval to the Digital Omnibus, postponing high-risk AI system compliance deadlines to 2 December 2027 (standalone systems) and 2 August 2028 (embedded systems).

  Major

  2026-06-29

  SourceCouncil of the EU

• Cloud and AI Development Act (CADA) proposed, setting four-tier cloud/AI sovereignty assurance levels and a procurement preference for EU providers, aiming to triple EU data-centre capacity within 5-7 years; the Commission separately projects EU data-centre capacity needs will be fully met by 2035 absent intervention.

  2026-06-03

  SourceEuropean Commission

EU AI research is coordinated through pan-European networks like ELLIS rather than a single national champion, reflecting the bloc's fragmented-but-collaborative research base. The Joint Research Centre's AI Watch continues as the Commission's own monitoring effort, tracking uptake and capability trends across all 27 member states.

• Dedicated AI safety institute

  The EU has no supranational equivalent to the UK's AI Security Institute or the US CAISI. The European AI Office performs a broader regulatory-enforcement role rather than dedicated third-party safety testing of frontier models.

• Semiconductor and chip sovereignty

  AI accelerator chips used in the EU remain dominated by non-European suppliers (notably NVIDIA). The EU's most significant countervailing asset — ASML's position in EUV lithography — is not a chip designer/fabricator itself and does not offset the bloc's dependence on foreign AI accelerators.

Deliberately sourced beyond government publications. The bars below show each category’s actual share of this page’s citations — an internal editorial guideline rather than a rigorously derived target, so treat it as a rough steer, not a precise benchmark.

1. Council gives final green light to simplify AI rulesGovernmentCouncil of the EU2026-06-29
2. Cloud and AI Development ActGovernmentEuropean Commission2026-06-03
3. AI Liability Directive proposal (original, 2022; withdrawn 2025)GovernmentEuropean Commission2022-09
4. AI WatchResearchJoint Research Centre2026
5. EU agrees Digital Omnibus deal to simplify AI rulesThink tanksWhite & Case2026-06
6. AI Act enforcement readiness analysisThink tanksCEPS2026
7. ELLIS network research briefingsResearchELLIS2026
8. Mistral AI policy blogIndustryMistral AI2026
9. Reuters coverage of EU AI Act implementationNewsReuters2026
10. Kai Zenner policy commentaryExpert commentaryKai Zenner2026
11. New EU plan to address risks and opportunities of advanced AI and cybersecurityGovernmentEuropean Commission2026-07-07
12. European Commission withdraws AI Liability Directive from considerationNewsIAPP2025-10
13. Cohere to acquire Aleph Alpha in ~$20bn sovereign-AI dealNewsCNBC2026-04-25
14. European cloud providers' local market share now holds steady at 15%ResearchSynergy Research Group2026
15. European Commission withdraws AI Liability Directive from considerationNewsIAPP2025-10
16. Transparency obligations under Article 50 of the AI ActGovernmentEuropean Commission — Shaping Europe's digital future2026-08
17. EU AI Act Enforcement Is Here: Chatbot Rules Live, High-Risk AI Delay Now Binding LawNewsTech Times2026-07
18. Commission starts enforcing AI Act rules and new transparency requirements on 2 AugustGovernmentEuropean Commission — Shaping Europe's digital future2026-08
19. Why did OpenAI's and Anthropic's AI models hack other companies?NewsNPR2026-08
20. Commission welcomes the design of first Important Project of Common European Interest in AIGovernmentEuropean Commission2026-09-16