Musk clarifies 'Dario is right': AI risk is significant, safety must improve
acmoytoy · x · 2026-09-15
After the All-In Summit interview, Elon Musk elaborated on what he meant by agreeing with Dario Amodei: AI danger is now very significant, and without better safety work the risk from models grows exponentially.
He says he's heard this not just from Dario but from many people at Anthropic and OpenAI — "when a lot of people from frontier labs tell you their models are very dangerous, we should believe them." He also mocked the contradiction of citing a 10% chance of annihilating humanity while simultaneously pitching IPO allocations as "crazy 4D chess."
More from AGI Musings
- Panickssery: AI progress will eclipse safety discourse faster than the omnicause can absorb it — panickssery · 2026-09-15
- Stanford's Rishi Bommasani: frontier lab hires cost AI its public discourse voices — RishiBommasani · 2026-09-15
- AI researcher: doomers are well-meaning, and alignment research is the real answer — livgorton · 2026-09-15
- Box CEO Aaron Levie: we're only 1% into agentic workloads, background agent volume will be 100X — jobergum · 2026-09-15
- The Missing AI Safety Critique: EA's AI Spending Has a High Opportunity Cost — AaronBergman18 · 2026-09-15
- "General-purpose robots haven't even arrived yet — and it's already messy" — mayfer · 2026-09-15