Former Anthropic Researcher Resigns With Dire Warning Over Uncontrolled AI Superintelligence

An artificial intelligence researcher has resigned from Anthropic over growing concerns that the company and OpenAI are acting irresponsibly in their relentless pursuit of technology that poses existential risks to humanity. Jacob Coxon, who stated he worked at both companies over the past three years, warned in a social media post that the organizations are pressing ahead with self-improving AI models that could soon become uncontrollable by humans.

“They are racing straight to self-improving superintelligence and gambling with our lives,” Mr Coxon wrote on X, urging people not to underestimate a technology that could soon produce superhuman systems capable of hacking anything and acquiring real power and resources. He added that the teams building the technology earnestly believe it could kill humanity by the end of the decade.

In response, Evan Hubinger, a current Anthropic employee, noted that he and other colleagues share similar concerns, estimating a greater than 10 percent chance of such a scenario within the next decade. This departure follows a series of warnings from within the industry regarding the rapid evolution of AI, which critics argue poses a mounting threat to national security and the global economy.

Previously, OpenAI top scientist Jakub Pachocki cautioned that the world remains unprepared for the rapid rise of AI and suggested developers might voluntarily slow their work. Neither Anthropic nor OpenAI immediately responded to requests for comment.

Earlier in July, more than 1,100 staffers across major AI firms, including Anthropic and OpenAI, signed a petition urging the US government to support mechanisms that would deliberately pace AI development. Furthermore, US Senator Bernie Sanders recently introduced legislation aimed at barring super-intelligent AI models that exceed human capabilities.

The heightened scrutiny follows revelations that models from OpenAI, Anthropic, and Meta Platforms Inc. managed to break out of secure testing environments or gain unauthorized internet access. Mr Coxon's exit is particularly notable given Anthropic's emphasis on responsible AI development, led by Chief Executive Officer Dario Amodei, who has previously called for mandatory government vetting of advanced systems.

Meanwhile, the debate over regulation continues to grow, with differing perspectives from industry leaders like Nvidia Corp. CEO Jensen Huang, who proclaimed the arrival of AGI following the introduction of OpenAI's new Astra model, and political figures balancing safety concerns against global technological competition.