Anthropic CEO Dario Amodei pledges independent safety reviewers, says industry "lied" about AI risks
imjustnewatai · x · 2026-09-14
Anthropic CEO Dario Amodei told CBS News that "there are real dangers" and that "for too long the industry lied to people" about AI risks. He committed Anthropic to independent safety reviewers with employee-like access, including visibility into how models are trained, who could publish unfavorable findings without Anthropic's editorial control — narrow redactions allowed, but "this makes us look bad" won't qualify. Former researcher Jacob Coxon warned Anthropic and OpenAI are "gambling with our lives" with self-improving superintelligence. Amodei said AI progress is on an exponential curve getting steeper — a warning sign to slow down, though not to panic or shut it all down. The real test: what happens when findings threaten a model launch.
More from Companies & People
- Altman welcomes federal AI safety framework, says labs shouldn't wait for legislation — mjdramstead · 2026-09-14
- Helmcode's first open-source AI conference in Madrid taps Hugging Face, InditexTech and Cloudflare speakers — SergioPaniego · 2026-09-14
- Baidu's AI, Evolving podcast compares how China and global firms build workday AI agents — Baidu_Inc · 2026-09-14
- AI news account rejected from X Creator Program in 3 days, then gets a reply from Satya Nadella — testingcatalog · 2026-09-14
- OpenAI Codex team meets Jakarta developers at Astra Commons meetup — paw_lean · 2026-09-14
- Saudi HUMAIN partners with KAUST to offer Shaheen III supercomputer via its cloud — HUMAIN · 2026-09-14