Former Anthropic researcher Jacob Coxon announced his resignation from the AI sector, declaring the race to build superintelligent systems out of control and urging immediate government intervention. He warned that the technology threatens humanity’s very existence.
Anthropic Researcher Quits AI Industry, Calls for Government Action
Coxon, who spent three years on pretraining research at OpenAI and Anthropic, said both companies are gambling with humanity’s future by racing toward self‑improving AI. He warned that the technology threatens humanity’s very existence.
“I resigned from Anthropic today,” he wrote on Twitter. “Neither company is acting responsibly; they are racing straight to self‑improving superintelligence and gambling with our lives.”
He warned that “these will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources.” He said the pace of progress shows no sign of slowing.
“The people building AI earnestly believe that it could kill us all by the end of the decade,” he said, citing private conversations with senior researchers. He emphasized that this is not a marketing stunt.
According to the Wall Street Journal, Coxon argues that safety trade‑offs are inevitable when firms compete, especially against Chinese upstarts. He says competition forces risky shortcuts.
Evan Hubinger, a colleague at Anthropic, echoed the concern: “I personally think the risk is greater than 10% within the next decade,” he told reporters. He noted that current alignment solutions are insufficient.
OpenAI has recently touted breakthroughs it calls a “generational leap” toward artificial general intelligence, with the GPT‑6 Astra model demonstrating major gains across science, coding, cybersecurity and professional work. The model’s capabilities represent a dramatic acceleration in AI performance.
Axios reported that OpenAI’s leadership and industry figures such as Nvidia CEO Jensen Huang have openly referenced the arrival of artificial general intelligence, defining it as AI that outperforms humans across most economically valuable work. Huang called the development a historic milestone.
In a blog post titled “An Alien Mind,” OpenAI researcher Pachocki warned that no lab has solved alignment and monitoring sufficiently to scale rapidly without compromising safety. He argued that unchecked scaling could outpace safety measures.
Coxon left OpenAI earlier this year, believing Anthropic was more responsible, but now says no company can responsibly develop superintelligent AI without government guardrails or a coordinated industry slowdown. He insists that voluntary measures have proven insufficient.
“The government has to step in and set hard and fast guardrails that everyone can follow,” he said, noting that voluntary measures have proven insufficient. He insists that only a top‑down executive order can establish enforceable standards.
Industry Debate Intensifies
OpenAI’s rapid rollout of GPT‑6 Astra has ignited an internal debate over how fast the AI race can proceed safely. Coxon’s resignation underscores growing alarm among leading researchers that the race to AGI may outstrip safety measures, demanding urgent policy response.
Call for Immediate Government Action
Coxon insists that only a top‑down executive order can establish enforceable standards, urging leaders to act before the technology reaches a point of no return. He argues that China’s pace does not excuse inaction, and that immediate steps are required to prevent a catastrophic outcome.
