A researcher at a big-time A.I. safety and research company says he’s quitting the biz because the industry is rushing to build systems that will spin out of control and destroy humanity … and he says the people building A.I. “earnestly believe it could kill us all by the end of the decade.”

Jacob Coxonquit his job as a researcher at Anthropic, where his specialty was training new A.I. models that consume vast amounts of data … and he’s sounding the alarm on his way out the door.

Coxon told the Wall Street Journal he decided to leave the company because he believes the AI systems he worked with will one day run amok and wipe out civilization.

He also said his colleagues are now using words such as “crunchtime” and “endgame” to characterize how they’ve fast-tracked the self-improving A.I. models.

And, get this … Coxon told the WSJ, “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.”

Coxon says he worked at OpenAI earlier this year before leaving to join Anthropic … and he claims competition between the major A.I. players is making the business — and humanity — less safe. He blasted both companies, saying they are “not acting responsibly.”

After the WSJ published their story, Coxon hopped on social media and posted more about his exit, his experience with OpenAI and Anthropic, and raised even more concerns.

Anthropic executives and other A.I. business leaders have constantly sounded the alarm bell, warning about the risks of A.I. models going off the rails. They’ve also recommended that the industry as a whole slow down development of the models.

Coxon is saying that isn’t happening, though … claiming, “They are racing straight to self-improving superintelligence and gambling with our lives.”