Anthropic Security Chief Argues for Dropping the 'Solve Alignment' Framing
Anthropic's Joshua Saxe argues AI safety should emulate how civilization manages complex systems rather than 'solving alignment,' a view echoed by others calling for layered defenses assuming agents are sometimes unsafe.
2026-11-20 ~ 2026-11-20 · 2 related posts
- Anthropic's Joshua Saxe: ditch "solving alignment" — AI safety must manage complex systems, not prove them — binarybits · 2026-11-20
- AI agents can never be proven safe—build layered defenses instead — binarybits · 2026-11-20