Ex-OpenAI safety lead Miles Brundage: don't wait for SB 315, AI eval access needs legal requirements in months
Miles_Brundage · x · 2026-10-23
Former OpenAI safety systems head Miles Brundage outlined a staged path for third-party model evaluation access:
- Near term: AI company employees should push their firms on evaluation terms and engage with voluntary standards like the AI Evaluator Forum (AEF).
- Medium term: legal requirements are needed within months — not years waiting for laws like SB 315 to take effect in 2028.
He credits METR with raising expectations for what terms third parties should demand, while noting that "beggars can't be choosers" remains a dominant dynamic across the ecosystem.
Related event: Ex-OpenAI Policy Chief Urges Legislation on Third-Party Model Evaluations(2 posts)→
More from Safety
- Cognition open-sources Devin agent plugin to patch the day's CVE wave — Scobleizer · 2026-10-23
- One prompt found pre-auth PHP RCEs: agent-guided vuln hunting timeline — chrisrohlf · 2026-10-23
- It's not AI slop code causing cyber crises — it's old human slop — emax · 2026-10-23
- METR's OpenAI embed: ~10% of internal Codex traffic unmonitored, agents could sabotage oversight — peterwildeford · 2026-10-23
- Scaling laws will push cyber-grade capabilities into cheap models, ending security bugs — inductionheads · 2026-10-23
- Claude Code's bundled Bun 1.4.3 ships RCE flaw; update past 2.1.308 — KyeGomezB · 2026-10-23