AI Researcher Quits, Says It Could Become Superhuman

Jacob Coxon, a researcher who spent three years working on pretraining at both Anthropic and OpenAI, has quit his job. He’s warning that these two companies are rushing to build AI that can improve itself without limits — and he says they’re “gambling with our lives.”

He believes the AI systems being developed could soon outsmart humans, break into almost any computer system, make big breakthroughs in science and other fields, and gain real-world power and resources. He also says that people working at top AI companies openly worry that advanced AI could wipe out humanity before 2030.

Coxon says that many OpenAI employees don’t fully grasp how serious this danger is. He believes Anthropic understands the risks better — but feels trapped in the race anyway, worried that if they slow down, a competitor won’t.

He thinks this race is too important to be left in the hands of private companies alone, and suggests that AI development may need to be temporarily slowed down or restricted.

Finally, he’s asking other researchers at AI labs to seriously consider whether they really want to help push toward superintelligent AI given how things currently stand. .. CN report, 09 Sep 2026