Anthropic researcher resigns, says AI companies ‘gambling with our lives’ and ‘it could kill us all by the end of the decade’

A researcher says he has resigned from one of the major AI companies and is now sounding a dramatic warning about the future. He alleges that both Anthropic and OpenAI are behaving irresponsibly by aiming for a “self-improving superintelligence” that he says has a real chance of wiping humanity off the face of the Earth.

Jacob Coxon, 27, is an AI researcher and math graduate who has worked in training new AI models at both OpenAI and Anthropic. Newsweek confirms he was a member of the OpenAI technical staff from 2023 to 2026, worked on the GPT-4o model, and joined Anthropic earlier this year to train their models.

Now Coxon is sounding the most explicit possible warning about what could soon be coming down the pipeline. In an X thread with almost 300k likes, he argues that these companies are “gambling with our lives” and underlines that we shouldn’t “underestimate the power of this technology.”

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

Coxon argues: “These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing.” He also argues that his opinion is widely shared within the AI tech community: “The people building AI earnestly believe that it could kill us all by the end of the decade.”

He goes on to anticipate the response to these claims: “if they truly believe this, why are they still building it?” He alleges OpenAI hasn’t “deeply internalized the civilizational stakes”, and that while Anthropic understands the dangers, “they are locked in a race to get there first – they believe no one else will act responsibly, so they must do it themselves, despite the risk.

He concludes: “If you are a lab researcher, I urge you to consider what the next few years will actually feel like. Do you want to kick off a superintelligent RL run without a rigorous understanding of its mind? Should you put your head down because “it’s happening anyway” – or take this moment to call for different conditions?”

Coxon’s conclusions were backed up by Evan Hubinger, Anthropic’s acting Alignment Science Lead (who also previously worked at OpenAI). He said, “Jacob is correct here,” and added: “We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.”

The general response in the replies is fear and confusion, together with people pointing out that maybe pursuing a course of action that might kill humanity isn’t a great idea. One said: “10% chance we’re going kill everyone within 10 years but idk man no one knows what to do fr”, another “Maybe you guys should think about simply not doing all this then?”, and, simply, “So just to be clear the work you’re choosing to do every single day has a greater than 10% chance of killing all humans. And you’re flippantly acknowledging this to everyone on social media. My goodness.”

While Hubinger is a current Anthropic employee, neither the company nor OpenAI has officially commented on the allegations that their technology has a chance of bringing about the extinction of humanity.