Andy Matuschak on AI Risk: Catastrophic Tail Risk Can't Be Ignored

Andy Matuschak argues that current AI risk debates talk past each other: even if things are 'probably fine,' the catastrophic tail risk remains significant, so we shouldn't roll the dice without control plans. The core disagreement is whether beneficial properties like reflectivity and corrigibility can reliably outpace dangerous capabilities.

2026-09-29 ~ 2026-09-29 · 2 related posts

• Andy Matuschak: don't roll the AI risk die without a real control plan — andy_matuschak · 2026-09-29
• Matuschak: you haven't shown safety properties develop fast enough — andy_matuschak · 2026-09-29