Synthesis history

4 versions, newest first.

Version 4 2026-07-22 02:13 UTC · 36 items

Two concrete developments this pass. First, OpenAI's CAISI proposal has advanced legislatively — NIST's AI Testing Office has been allocated $20M [^41253] — giving the previously paper-stage proposal real funding tracti…
Version 3 2026-07-21 02:06 UTC · 30 items

Three substantive additions this pass. First, OpenAI moved from expressing support to proposing its own distinct mandatory federal mechanism — CAISI — with annual audits and incident reporting, creating a concrete polic…
Version 2 2026-07-16 18:23 UTC · 21 items

Two substantive additions this pass. First, observers identified similarity between Hassabis's proposal and an earlier proposal by Mark Thomas published at Lawfare [^40885], establishing intellectual precedent and sugge…
Version 1 2026-07-15 18:10 UTC · 15 items

Demis Hassabis, CEO of Google DeepMind, has published a proposal for a U.S.-led self-regulatory body modeled on FINRA to test frontier AI models for cyber, biological, and deception risks before public release [^40724][…