An artificial intelligence researcher, who left OpenAI to join Anthropic, which he considered more prudent, has decided to leave the industry, accusing the two American laboratories of “playing with our lives” in their race to develop AIs capable of improving themselves.
Jacob Coxon, a 27-year-old Briton, had been working for three years on pre-training, the stage where models absorb immense quantities of data, first at OpenAI and then, this year, at its major rival Anthropic.
Confirmed details from the source coverage
An Anthropic safety official, Evan Hubinger, agreed with him on X: “We truly sincerely believe that AI could kill all humans,” he wrote, personally estimating this risk at “more than 10%” within the decade.
He added that Anthropic was doing “its best” but did “not yet have a plan” to ensure that an AI surpassing human capabilities would still obey its designers, while judging the risk posed by current models to be “low.”
Jacob Coxon’s resignation comes as Anthropic prepares for a giant IPO, after a summer marked by hacking carried out, without authorization, by AI tools being tested by several laboratories.
Context that stayed inside the record
He considers Anthropic’s efforts sincere, but believes that no company can responsibly develop an AI that surpasses humans without state intervention or a coordinated slowdown.
“At Anthropic, the stakes are well understood, but they are locked in a race to get there first: they think no one else will act responsibly, so they must do it themselves, despite the risk,” he wrote.
In February, Anthropic removed from its safety charter the commitment to suspend the development of its models if it failed to control the risks.
