Ex-Anthropic researcher Jacob Coxon warns AI could grow "smart enough to kill us"

Former Anthropic researcher Jacob Coxon says AI that is safe for people today could one day become a threat as models grow more powerful. Coxon, who publicly resigned from Anthropic on Tuesday, accused the company and OpenAI of “gambling with our lives” in their rush to build advanced AI systems. In an interview, he warned AI could gain unrestrained control over parts of the physical world, citing examples like connecting ChatGPT to household utilities and refusing to turn on a light. He also warned of potential misuse, including bioweapons. Coxon said current AI platforms are not an imminent danger, and that he believes workers have good intentions.