A researcher has publicly resigned from the prominent artificial intelligence startup Anthropic, issuing a stark warning that leading AI companies are moving too quickly and recklessly toward potentially uncontrollable AI systems.
Jacob Coxon, 27, announced his departure in a viral thread on X (formerly Twitter). Coxon, who spent the past three years conducting pretraining research at both OpenAI and Anthropic, stated that he is leaving the AI industry altogether due to profound ethical and safety concerns.
“Neither company is acting responsibly,” Coxon wrote. “They are racing straight to self-improving superintelligence and gambling with our lives.”
The “Endgame” of AI Development
Coxon’s resignation highlights a growing rift in the AI industry over “self-improving superintelligence”—the hypothetical point at which AI systems become capable of substantially improving their own code and capabilities without human oversight. According to Coxon, these advancements could soon yield systems capable of hacking critical infrastructure, revolutionizing fields overnight, and acquiring significant real-world power and resources.
More alarmingly, Coxon claims that the fear of a catastrophic outcome is widespread within the walls of top AI labs, even if it is downplayed in public. “The people building AI earnestly believe that it could kill us all by the end of the decade,” he warned. “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.”
A Dangerous Race
Coxon also shed light on the intense competitive dynamics fueling this rapid development. While he noted that concerns about AI’s long-term risks are generally better understood at Anthropic than at his previous employer, OpenAI, he argued that this awareness hasn’t slowed development.
He claimed that Anthropic is locked in a fierce race to achieve advanced AI first, driven by the belief that if they don’t, a less responsible organization will. Coxon characterized this competitive mindset as a “hubristic gamble.”
Coxon’s departure comes as Anthropic prepares for a reported $2 trillion valuation IPO and follows a recent push by over 1,000 industry researchers and executives—including Anthropic CEO Dario Amodei—urging governments to create a “brake pedal” for AI development. While Coxon remains optimistic that coordinated action and pacing agreements between U.S. labs are still possible, he argues that stronger interventions, such as a temporary ban on improving model capabilities, may ultimately be necessary to prevent a global disaster.


