According to leading media news outlets, an Anthropic researcher is quitting the artificial intelligence industry entirely. Allegedly, he cited fears that leading labs are racing to build systems they cannot control.
The researcher in focus is Jacob Coxon, aged 27. He spent three years doing pretraining research at both OpenAI and Anthropic. Pretraining involves training new AI models on vast amounts of data. Coxon announced his resignation through a series of posts on X.
His warning was strikingly blunt and direct. He wrote that neither company is acting responsibly currently. Instead, he said they are racing straight toward self-improving superintelligence. Consequently, he accused the industry of “gambling with our lives.”
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
— Jacob Coxon (@hilbertspaess) September 9, 2026
Coxon’s reportedly told media that aggressive scenarios could arrive soon, specifically, he warned things could be “out of control” by late next year. His central fear centers on recursive self-improvement directly.
“We’re on track for a lot of the most aggressive of these scenarios where by the end of next year things could be out of control already,” he said.
To those unfamiliar, this concept describes AI improving its own capabilities repeatedly. Essentially, the AI takes over enough research to accelerate its successors. As a result, development could outpace researchers’ ability to evaluate it. Coxon warned future systems could hack anything and acquire real power.
Anthropic also seems to at least have a cognizant acknowledgment of dampning such innovation. In theri own words:
If it were possible to effectively slow the development of this technology to give ourselves more time to deal with its immense implications, we think that would likely be a good thing. But if a slowdown simply lets the least cautious actors catch up technologically, it could leave everyone less safe. Without a global coordination mechanism, companies and governments will have to make difficult decisions about safety while under competitive and geopolitical pressures.
Notably, Coxon joined Anthropic specifically for its safety reputation. He does not claim the company’s safety work is fake. Rather, he argues competition makes those safeguards fundamentally insufficient, partly because, as he believes, no single lab can safely win this race alone.
He also drew a distinction between the two labs. At OpenAI, he said many haven’t internalized the civilizational stakes. At Anthropic, the risks are understood but the race feels unavoidable. The company believes it must arrive first to act responsibly.
Coxon believes that the issue has grown beyond the scope of any single company. He suggests that we might need some serious collaboration across the industry or even with the government before self-improving AI systems become too difficult to manage.
As WSJ put it:
Coxon said he is leaving the company because he doesn’t want to be part of an industrywide race to build AI systems capable of improving themselves, fearing they could eventually spiral out of control and threaten humanity.
