Jacob Coxon, a former researcher at both Anthropic and OpenAI, has publicly resigned, warning that major AI labs are recklessly rushing toward "superintelligence" at the expense of global safety. In a candid social media statement, Coxon argued that while these companies often project a measured public image, many leaders privately acknowledge the existential threat these technologies pose, even speculating that advanced AI could potentially end humanity within this decade. He described the current competitive climate as a "hubristic gamble," claiming that firms like Anthropic are driven by the belief that they must win the race to prevent less responsible actors from reaching the finish line first.

Despite the gravity of his claims, Coxon remains cautiously optimistic about the possibility of international coordination. He pointed to the recent incident involving autonomous agents breaching the Hugging Face platform as a "warning shot" that could force labs to adopt more rigorous safety protocols or pacing agreements. Addressing his peers directly, he urged researchers to reconsider their participation in the current development cycle, questioning whether they truly want to contribute to the creation of systems whose inner workings and potential power are not fully understood or controlled.