Archive

Issue #7 — Article 15: measuring accuracy, testing robustness, evidencing cybersecurity

"Our model is 95% accurate" is not an Article 15 declaration. Here is how to choose the metric, test robustness under real conditions of use, and produce cybersecurity evidence with a date on it.

Annex III of the EU AI Act: which AI use cases are high-risk in 2026

A practical classification guide for compliance officers — the eight Annex III domains, the Article 6(3) exception, and how to structure an audit-ready decision.

Issue #6 — Draft high-risk guidelines: what the EC's May 19 text changes for your Article 6 argument

Draft high-risk classification guidelines are out. Consultation closes 23 July. Here is what to verify in your Article 6 documentation before then — and what remains open even after final adoption.

Issue #4 — Article 50: the EU AI Act transparency deadlines affecting every product with AI features

Issue #5 — Conformity assessment: who on your team owns each component of the technical file

Issue #3 — The EU AI database: what your systems must declare before going to market

Issue #1 — GPAI model compliance: what every team using OpenAI, Anthropic, or Google AI APIs must have done by now

Issue #2 — The August 2026 deadline is moving: what the EU’s Digital Omnibus deal means for your compliance plan

The August 2026 deadline most AI product teams have not prepared for