Definition
The EU AI Act is the European Union’s risk-based regulatory framework for artificial intelligence systems, imposing obligations by risk tier on providers and deployers.
Key Points
- 2026-10-05: openai deploys textgrain watermarking EU-only for chatgpt/codex — first major US lab regional rollout; API opt-in globally (2026-10-05-openai-chatgpt-watermarking-eu)
- 2026-08-02: Article 50 transparency obligations applicable — generative outputs must be machine-readably marked where technically feasible
- 2026-07-31: Code of Practice on Transparency published; ~190 signatories including anthropic, google, openai, Meta, Microsoft, Mistral
- 2026-08-14: anthropic implements synthid-text watermarking globally on future Claude models (2026-08-14-anthropic-claude-text-watermark)
- Target mapping framework for asago open-source governance orchestration
- Compliance pressure cited in commercial agent-data governance launches (e.g. ethyca astralis)
- Interacts with U.S. voluntary approaches (ai-cybersecurity-testing) as dual-regime enterprise reality
Related
- textgrain
- openai
- synthid-text
- content-watermarking
- anthropic
- ai-governance
- ai-regulation
- nist-ai-rmf
- asago
- ethyca
- ai-governance-orchestration