Anthropic Researcher Resigns, Warns AI Labs Privately Fear It Could ‘Kill Us All'

An Anthropic researcher, Jacob Coxon, said he resigned from the company, alleging that Anthropic and OpenAI are racing toward self-improving superintelligence and gambling with human lives.

The researcher said he spent the last three years on pre-training work at both companies. He made the claims in a series of posts on X.

Anthropic Researcher Walks Out With Grim Warning For Everyone Still Inside

Coxon argued that the technology’s trajectory is widely underestimated. He claimed that coming systems will hack anything, transform any field overnight, and acquire real power and resources.

He also alleged that private conversations inside the industry differ sharply from public messaging. According to Coxon, senior figures soften their phrasing for the press while expressing fear privately.

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

Coxon drew a distinction between his two former employers. He claimed many at OpenAI have not internalized what he called the civilizational stakes, while Anthropic understands them but stays locked in a race to arrive first.

Jacob is correct here—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. https://t.co/QAIHiFP3QZ

— Evan Hubinger (@EvanHub) September 9, 2026

He is one of the many employees to break ranks this year. Anthropic’s Safeguards Lead, Mrinank Sharma, resigned in February.

Joshua Achiam, formerly OpenAI’s chief futurist, wrote on September 1 that rogue AI systems will replicate in the wild and pursue money and power.

Follow us on X to get the latest news as it happens

The Call For Coordination

Coxon described entering what he termed the endgame as a hubristic gamble. He argued such a decision should not be launched from a private company’s Slack.

However, he said he remains optimistic about coordination. He claimed the Hugging Face attack has made pacing agreements between US labs more viable.

Both companies have already backed a version of that idea. More than 1,100 frontier lab staff signed the “Pacing theFrontier” letter in July, including Anthropic CEO Dario Amodei and OpenAI chief scientist Jakub Pachocki. 

“We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development,” the letter reads.

In addition, both companies have also signed a joint open letter on cyber defense, alongside Google, Microsoft, and roughly 150 other organizations. The letter warns that AI-enabled attacks will become far more widespread and sophisticated in the coming months.

Coxon’s warning also lands shortly after UN human rights chief Volker Türk told the Human Rights Council that advanced AI could pose an existential risk.

Coxon added that preventing a global race may require a temporary ban on efforts to improve model capabilities. He closed by urging lab researchers to consider whether they want to start a superintelligent reinforcement-learning run without understanding the resulting system.

BeInCrypto has reached out to Anthropic and OpenAI for comment.

Subscribe to our YouTube channel to watch leaders and journalists provide expert insights