Standing
The AI Act's safety requirements are not settled in the Regulation. They are settled in the drafting of a handful of technical standards, in committee rooms most readers never see into. The reason to read this Observatory is that the people writing it are in those rooms.
The contributors behind these briefs also help write the standards the briefs read. Across the collaborators that means active roles — experts, convenors and project leaders — inside CEN/CENELEC JTC 21, the committee producing the AI Act's harmonised standards, and in the international work at ISO/IEC JTC 1/SC 42 that the European frame draws on and adopts from.
Participation is in an expert capacity. The readings on this site are the Observatory's own; they represent no committee, national body or institution. That independence is the point: proximity to the drafting, distance from the verdict.
The joint technical committee drafting the AI Act's harmonised standards under mandate M/613 — the prEN 18229 trustworthiness family, the Article 17 quality-management standard, risk, datasets, and the rest of the programme this series reads.
The international committee for artificial intelligence. Much of what SC 42 produces is drawn into the European frame by reference or adoption — which is why the outer ring of the mark reaches past Europe's edge.
Companion resource
Where the Observatory reads what a clause does, a companion project maps where each standard sits: which CEN-CENELEC project covers which article, which carry an Annex Z citation, and how far each has moved through the drafting stages. The interactive AI Act standards map is built by Adam Leon Smith (Piccadilly Labs) and supported by Trustora Digital. Take it as the atlas; take these briefs as the field notes.
Open the standards map →The organisations behind the Observatory are listed on the collaborators strip, and set out with their roles on the About page. How the readings are made is stated briefly in the method.