AI Insiders Issue Terrifying Warning to Humanity

Three researchers connected to artificial intelligence giant Anthropic are sounding the alarm over what they believe could become one of the greatest threats humanity has ever faced.

The researchers have warned that rapidly advancing AI systems could potentially wipe out the human race before the end of the decade — with some insiders reportedly fearing catastrophe could arrive within just a few years.

Jacob Coxon, who recently resigned from Anthropic after spending three years working at Anthropic and OpenAI, claimed many of the people actually building the world’s most powerful AI systems privately believe the technology could become deadly.

“The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon wrote on X.

He insisted the warnings are not some kind of publicity stunt, claiming researchers and executives sometimes soften their language when speaking publicly despite expressing much greater concern behind closed doors.

“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 said.

He added: “No other human activity poses this level of danger.”

Coxon also accused both Anthropic and OpenAI of failing to take the threat seriously enough, saying that “neither company is acting responsibly.”

His concerns were publicly backed by Anthropic alignment researcher Evan Hubinger, whose work focuses on ensuring increasingly advanced AI behaves in ways that remain compatible with human interests.

“We really do earnestly believe AI could kill all humans!” Hubinger said.

Hubinger estimated there is a greater than 10 percent chance of such an outcome over the next decade.

He also acknowledged that Anthropic still does not have a reliable solution for one of the biggest unanswered questions surrounding advanced AI — how to guarantee a system far smarter than humans will continue following human goals and values.

Another Anthropic researcher, Samuel Marks, who works on scalable oversight, offered an equally disturbing assessment.

Marks said some of the people developing the technology believe human extinction could potentially occur within “the next few years.”

The extraordinary warnings come as AI companies race to build increasingly capable systems, raising new questions about whether developers will be able to control the technology if it eventually surpasses human intelligence.

For researchers like Coxon, Hubinger and Marks, the concern is no longer simply that AI could eliminate jobs, spread misinformation or disrupt society.

Their fear is much bigger — that humanity could eventually create a system powerful enough to destroy the people who built it.