A former Anthropic employee has sparked fresh fears about artificial intelligence, Jacob Coxon quit his job and didn’t hold back on his way out.
The 27-year-old researcher shared his concerns in a resignation thread on X.
This Anthropic AI safety warning painted a troubling picture of the industry, Coxon claimed AI systems could soon become dangerously powerful.
He described what’s coming as superhuman technology tphese systems could “hack anything, revolutionize any field overnight, and acquire real power and resources.”
Coxon didn’t mince words about the industry’s direction. “They are racing straight to self-improving superintelligence and gambling with our lives,” he posted.
His warning isn’t happening in isolation and a growing number of AI workers have raised similar concerns in recent months.
In July, nearly 1,400 AI employees signed a letter pushing for government regulation, they want tighter control over how fast AI develops.
Senior Anthropic researcher Evan Hubinger backed up Coxon’s claims publicly.
“We really do earnestly believe AI could kill all humans!” he posted.
Hubinger estimated the risk at under 10% over the next decade. He later clarified that Anthropic’s own risk report says current AI poses little such danger.
Anthropic responded with an official statement addressing the concerns.
“We have always been transparent that AI will bring both enormous benefits and unprecedented risks,” a company spokesperson said.
This Anthropic AI safety warning adds to wider industry unease CEO Dario Amodei has also warned about the risks of moving too fast.
Meanwhile, US regulation remains slow to catch up; the White House has floated a voluntary review framework.
Anthropic says it wants clearer industry-wide rules.
“We believe the world would benefit from the industry adopting a lawful, verifiable way to work together to pace how we release powerful models,” the company said.
