Richard Ngo: No Need to Stay in Top Labs for Alignment Work

RichardMCNgo · x · 2026-08-28

Richard Ngo advises employees at top labs like Anthropic and OpenAI: 1. You don't need to stay to do research, as work on scaling intelligence can usually be done on public models. 2. You don't need to stay for influence, as fear prevents wielding it effectively. 3. You don't need to stay to generate scary demos, as warning shots will be abundant. Dwight Kokotajlo endorsed this, noting Richard's conversion from an insider dismissive of such views adds credibility.

Related event: Richard Ngo: You Don't Need to Stay at Top Labs for AI Safety Research(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →