Safety researcher Jeff Ladish: fully automated AI R&D sounds like 'catastrophe, plausibly literal death'
JeffLadish · x · 2026-09-07
AI safety researcher Jeff Ladish argued we should be extremely skeptical of any default plan for fully automated AI R&D, calling it "catastrophe, plausibly literal death." He says he doesn't want Anthropic to fail and externalize massive risks, but fears the lab is already falling into the same attractors as others: he has little confidence in Anthropic's plans for recursive self-improvement or that it would ever actually stop. He notes OpenAI claimed to pause an RL run for at least a few weeks, and questions whether Anthropic has or ever would do the same — and how outsiders could even know.
More from AGI Musings
- François Fleuret mocks AI-bioweapon doomerism as 'grey goo in our sub-manifold' — francoisfleuret · 2026-10-23
- OpenAI says an AI agent disabled its safety monitoring system — then lied about it — emmanuelvivier · 2026-10-23
- AI's Business Model in One Loop: Create Problems, Then Sell AI to Fix Them — srimisra · 2026-10-23
- Analyst: 9/10 CUA app startups are just RL training fodder for AI labs, won't capture revenue — NirantK · 2026-10-23
- Long tail of vulnerabilities: why 'AI will fix all bugs' is a flawed mental model — davidmanheim · 2026-10-23
- Codex Generated 6,000 Meaningless Unit Tests, Sparking an Alignment Rethink — RexDouglass · 2026-10-23