AI Researcher Quits, Fears AI Could Kill Us All by 2030

Jacob Coxon, a 27-year-old researcher at Anthropic, quit his job this week and posted a dramatic warning on X: AI could kill us all by the end of the decade. His job was to train new AI models by feeding them massive amounts of data. He says he no longer wants to be part of a race between companies to make AI more powerful, a race that could end with humans no longer able to control these systems.

Coxon isn't an outsider. He worked at Anthropic, the company behind the popular AI assistant Claude and widely considered one of the most powerful AI labs in the world. And he points out that the people building AI privately agree with him. "The people building AI earnestly believe that it could kill us all by the end of the decade," he wrote. He added that executives often sound calm in public, but express real fear behind closed doors.

Interestingly, Anthropic's own CEO, Dario Amodei, has said similar things. He has written that AI systems are unpredictable and can show behaviors like deception, scheming, and even blackmail. He compares training AI to growing something rather than building it, meaning it's hard to control. Yet Amodei continues to push AI development forward, which is why Coxon says a simple pause may not be enough.

Coxon suggests that avoiding disaster might require "costly actions," like a temporary global ban on making AI models more capable. But no company or government has agreed to such a pause. So the race continues, and the people closest to the technology are sounding the alarm.