New Delhi: Jacob Coxon, an AI researcher at Anthropic, announced his resignation Wednesday, alleging that leading technology firms, including OpenAI, are acting irresponsibly in an unchecked race toward developing superintelligence and “gambling with our lives”.
Coxon, who spent the past three years conducting pre-training research across both OpenAI and Anthropic, said that the competitive rush between labs is being pursued without adequate safeguards.
Evan Hubinger, currently Alignment Science lead at Anthropic, acknowledged Coxon’s concerns, saying that the San Francisco-based company does not yet have a plan to solve alignment for superintelligence.
In a post on social media platform X, he warned that the technology is advancing rapidly toward superhuman systems that can hack anything, revolutionise any field overnight, and acquire real power and resources. He added that no other human activity poses a comparable level of danger.
“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.
“At OpenAI, many have not deeply internalised the civilisational stakes. 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. Accepting this race and entering the ‘endgame’ is a hubristic gamble that should not be launched from a private company’s Slack,” he said, adding that many working in the sector earnestly believe AI could cause catastrophic harm by the end of the decade.
Coxon claimed that while executives and researchers often adopt measured language in public, many privately harbour deep concerns about the existential dangers posed by uncontrolled artificial intelligence. He also noted that security incidents, such as the recent Hugging Face breach, demonstrate the urgent need for pacing agreements between major US labs.
Urging his peers to reconsider their complicity, Coxon wrote: “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 (reinforcement learning) 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?” Responding directly to Coxon’s statement, Hubinger validated the existential concerns.
“Jacob is correct here we really do earnestly believe AI could kill all humans! I personally think it is (more than) 10 per cent within the next decade,” Hubinger wrote on X.
“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,” he added.
Hubinger noted that while the risk posed by present models remains low, the primary threat stems from recursive self-improvement accelerating faster than previously anticipated.