What we watch, and how often.

The AI Act itself, every act that amends or is based on it, corrigenda, implementing and delegated regulations.

This is where a change becomes law rather than an announcement. Regulation (EU) 2026/1744, which moved the high-risk deadlines, appeared here first.

A direct string match on the CELEX literal returns nothing against this store, which is why the adapter uses CONTAINS. Worth knowing if you query it yourself.

Guidance, templates, consultations and enforcement announcements from the body that supervises GPAI providers.

Most of the interpretive weight sits here. The text of the regulation rarely changes; the way it is read changes often.

The Commission's own help channel: FAQ updates, revised guidance on high-risk use cases, practical clarifications.

Every item on this feed is AI Act business by definition, so nothing is filtered out. Highest signal per item of all our sources.

Broader digital policy announcements, filtered down to those touching the AI Act, the AI Office, GPAI or the AI Pact.

Catches enforcement and policy news that never reaches EUR-Lex, such as the announcement that enforcement powers took effect.

Opinions and recommendations of the body that coordinates national authorities.

Where the member states settle on a common reading before it becomes national practice.

Drafts, the final code, signatory lists and subsequent revisions.

Signing the code creates a presumption of conformity for GPAI providers, so its wording has direct legal effect.

Opinions and guidelines where data protection meets AI, including the interplay between the GDPR and the AI Act.

In practice a data protection authority is often the first regulator an AI product meets, well before any AI Act enforcement.

Harmonised standards under the AI Act: work programme, drafts in enquiry, published standards.

A harmonised standard cited in the Official Journal creates a presumption of conformity. None has been cited yet, which is itself the news.

Announcements of the German market surveillance authority, which took that role when the KI-MIG entered into force on 29 July 2026.

If you place AI on the German market, this is the authority that inspects you. A global tracker does not carry this layer.

Positions of the German federal data protection commissioner on AI supervision.

Germany splits AI oversight across several bodies, and the boundaries are still being drawn.

Guidance on AI security, which feeds the robustness and cybersecurity duties under Article 15.

Article 15 says what to achieve, not how. This is where the how gets written down in Germany.

Recommendations and enforcement of the French data protection authority on AI systems.

The CNIL publishes early and in detail, often before an EU-level position exists.

The Dutch public register of algorithms used by government bodies.

A working example of the transparency the AI Act asks for, running ahead of the obligation.