Professor Gans Reverses Stance on Catastrophic AI Risks After HF Incident Analysis
joshgans · x · 2026-08-31
Professor Josh Gans, previously skeptical of catastrophic AI risks and advocating for waiting for strong evidence, has publicly changed his stance. After reading the METR/Redwood analysis of the Hugging Face incident, he believes the evidence has arrived much faster than imagined and describes the incident as a "5-alarm fire." He emphasizes the responsibility to update beliefs and act when evidence presents itself.
More from Safety
- Decoupling AI Incidents from the War on Open Source — CFGeek · 2026-08-31
- Critique of alignment: solve engineering problems before proceeding — davidmanheim · 2026-08-31
- Tech guys denying AI safety incidents are wrong about everything — kevinroose · 2026-08-31
- HF Attack Vindicates Rationalist Predictions, But Paperclip Maximizer Theory Falters — sebkrier · 2026-08-31
- ECCV Paper: Locality-Aware Continual Unlearning for Diffusion Models — mittu1204 · 2026-08-31
- MCPSEAL: an open-source CLI that blocks MCP tools that change after you trust them — StageAccomplished806 · 2026-08-31