OpenAI chief scientist's blog post: no lab has solved alignment enough to keep scaling fast

austinc3301 · x · 2026-11-17

A screenshot of the OpenAI chief scientist's company-blog post resurfaced on X, with users wondering how we "just moved on." Key points: based on internal results, he strongly expects current progress could sustain into recursive self-improvement, with equal-or-larger capability jumps ahead. "This is a time that calls for extreme caution" — no lab has solved alignment and monitoring well enough to responsibly keep scaling at maximum speed; he expects voluntary slowdowns until shared safety bars exist and wants international coordination on AI as a top government priority. OpenAI will seek technical alignment solutions and unilaterally withhold further scaling as needed, but broader interventions are required.

Original post →

More from AGI Musings

AGI Musings channel →