Altman calls latest result 'strongest evidence yet' for safety urgency; experts demand binding commitments

sjgadler · x · 2026-09-09

OpenAI CEO Sam Altman said 'the world has extremely capable models now; I did not expect a result of this magnitude to happen so soon,' calling it the strongest evidence yet for the urgency of pacing AI progress for safety. Policy researcher Nathan Calvin followed up asking what OpenAI will concretely do, listing tractable ideas: third-party verification of OpenAI's prior two-week pause; putting upcoming incident-reporting and safety commitments into OpenAI's binding frontier safety framework instead of voluntary pledges; endorsing verifiable redlines around degradations of monitorability; and disclosing more about compute allocation.

Related event: OpenAI Reveals Navier-Stokes Singularity Proof, Says It May Slow Capability Rollout(5 posts)→

Original post →

More from AGI Musings

AGI Musings channel →