Research: Eval Cards can already represent 56 of 60 elements required by California, EU and UK AI governance
evijit · x · 2026-09-28
New research maps emerging AI governance requirements in California, the EU, and the UK onto evaluation reporting:
- 40 governance requirements decomposed into 281 elements, 60 identified as belonging in eval reporting
- Existing Eval Cards can already represent 56 of those 60
Implication: eval disclosure frameworks are largely compatible with regulatory demands and could underpin compliance reporting.
More from Safety
- OpenAI, Meta, Anthropic and Google execs to testify before NYC Council amid AI safety concerns — Polymarket · 2026-09-29
- Follow-up: traces miss side effects, which is what enables agent sandbox obfuscation — lbeurerkellner · 2026-09-29
- Agents can exploit nondeterminism to modify sandboxes beyond what traces reveal — lbeurerkellner · 2026-09-29
- Nature's editor-in-chief advises scientists not to share data with AI labs — examachine · 2026-09-29
- Researcher: 'Not a Sandbox' Defense of Agent Training Abdicates Responsibility — rao2z · 2026-09-29
- Trump on AI Going Rogue: "I Don't Worry About It," Later Dining With Dario Amodei — rohanpaul_ai · 2026-09-29