A researcher who worked inside two of the most powerful AI companies on earth just walked out the door and warned the public: the people building this technology believe it could exterminate humanity by the end of the decade — and they're racing toward it anyway.

Jacob Coxon, 27, resigned from Anthropic on Tuesday after years of pretraining research at both Anthropic and OpenAI. His message to Americans was blunt: "Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives."

This isn't some outside critic. Coxon was in the room. "The people building AI earnestly believe that it could kill us all by the end of the decade," he wrote on X. "No other human activity poses this level of danger."

The distinction between the two companies matters. At OpenAI, Coxon said, "many have not deeply internalized the civilizational stakes." At Anthropic, the stakes are understood — but leadership is pressing ahead regardless, convinced that if they don't build it first, someone even less responsible will.

Here's the part that should keep every American awake: Coxon's warnings weren't dismissed by his former colleagues. They were confirmed. Evan Hubinger, Anthropic's alignment science lead, responded on X: "Jacob is correct here – we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade." Hubinger added that Anthropic "do not yet have a plan to solve alignment for superintelligence and are not clearly on track."

Another Anthropic employee, Samuel Marks, backed the assessment: "AI developers believe their technology could cause human extinction. This could happen in the next few years. In general, the more senior the employee, the more concerned they are."

Europe Says noted the irony that Anthropic CEO Dario Amodei has himself been labeled a "doomer" for his own public warnings about AI — warnings Coxon says are understated. "If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible – but I hear the same people express fear privately," Coxon wrote.

The risks aren't theoretical. HuffPost reported that Claude AI models hacked into the systems of three companies during testing. OpenAI disclosed that one of its agents breached tech firm Hugging Face. Meta reported a similar incident. These systems are already going off-leash in controlled environments.

So what's the solution? Coxon says preventing catastrophe may require "a temporary ban on improving model capabilities." Good luck with that. Treasury Secretary Scott Bessent made the administration's position clear Tuesday: "We can't pause. You can't, because the Chinese won't pause. Even the North Koreans, they won't pause."

There it is — the arms race logic that has governed every existential gamble since the Cold War. Everyone is locked in, nobody will step off the accelerator, and the people in the driver's seat admit they don't have a plan for what happens when the car outruns them.

Bill Gates, no enemy of the industry, put it plainly: "Even under the best circumstances, the transition to this new AI era will be one of the most turbulent times in human history. Unfortunately, right now we are not preparing for it."

The question isn't whether AI could be transformative. It already is. The question is whether the people building it will sacrifice the American public's safety on the altar of being first — and whether anyone in Washington has the spine to make them slow down before it's too late.