Anthropic researcher quits, warns AI labs are risking lives

Jacob Coxon, who has worked on pre-training research at both Anthropic and OpenAI, announced his departure on social media.

He said he had spent the past three years researching how new AI models are trained using very large amounts of data.

Coxon said his decision was linked to concerns about the direction of the AI industry and the speed at which companies are developing increasingly powerful systems.

Researcher warns of an accelerating AI race

In a thread on X, Coxon said he had decided to leave Anthropic after working in AI research for several years.

“I resigned from Anthropic today. I spent the last three years doing pre-training research at both OpenAI and Anthropic.

“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” he added.

Coxon argued that the capabilities of advanced AI systems could soon go far beyond what people currently expect.

“Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources,” he said.

“We have all witnessed the progress in each of these domains, and progress is not slowing,” he added.

His comments reflect a wider debate inside the technology industry about how quickly AI systems should be developed and deployed.

Some researchers believe rapid progress could bring major benefits, while others are increasingly concerned about systems becoming too powerful to control.

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

— Jacob Coxon (@hilbertspaess) September 9, 2026

AI safety concerns raised by former researcher

Coxon also claimed that some people working at leading AI companies privately fear that the technology could have extremely serious consequences.

“The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. 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. No other human activity poses this level of danger,” he said.

He questioned why companies would continue building increasingly advanced AI if their own researchers and executives have concerns about possible dangers.

“A common response is ‘if they truly believe this, why are they still building it?’ At OpenAI, many have not deeply internalised the civilisational stakes,” he said.

“At Anthropic, the stakes are well-understood, but 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 added.

The comments point to one of the biggest questions surrounding the current AI boom.

Companies are competing to build more capable systems, but researchers continue to debate whether safety measures are advancing quickly enough alongside those capabilities.

Also Read:PTA to block WhatsApp accounts linked to inactive SIMs

Concern over self-improving AI systems

Coxon urged other AI researchers to think carefully about what the next stage of development could mean.

“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?” he asked.

Speaking to the Journal, Coxon reportedly gave similar reasons for leaving Anthropic.

He said he did not want to take part in an industrywide rush to “build AI systems that can improve themselves, worried such systems could spiral out of control and destroy humanity”.

According to the report, Coxon said some of his colleagues in the industry had begun using terms such as “crunch time” and “endgame” when discussing the development of self-improving AI models.

Those descriptions suggest that some researchers view the current period as a critical stage in the race to develop increasingly autonomous systems.

The debate is not limited to one company, as concerns about AI safety have spread across research labs, technology companies and governments.

More researchers call for caution

Coxon was also among more than 1,000 AI researchers who recently signed a statement calling for greater international cooperation on AI development.

The statement urged governments around the world to coordinate efforts and create mechanisms that could “slow AI development if a brake pedal is needed to control models capable of improving on their own”.

The call reflects growing concern that individual companies may face strong pressure to move quickly because competitors are doing the same.

If one laboratory slows down while others continue developing more powerful systems, companies may fear losing their position in the rapidly changing AI market.

That competitive pressure is one reason safety researchers have increasingly called for common rules and international cooperation.

At the same time, supporters of continued AI development argue that the technology could improve medicine, science, productivity and many other areas.

The disagreement therefore centres not only on whether AI should continue developing, but also on how quickly that development should happen and what safeguards should be in place.