Several former engineers and directors from AI companies have resigned in recent months — burnout?
Jacob Coxon, formerly a researcher at "Anthropic", says that AI company executives privately fear a catastrophe, yet continue racing to build it anyway.
Jacob’s departure comes after a summer of failures in AI containment and a new initiative in Congress to completely ban superintelligent AI.
Jacob spent three years building the systems behind some of the world’s most powerful AI models and has now abandoned the sector entirely, warning that companies competing to develop increasingly capable AI “are playing with our lives”.
Jacob Coxon, 27, resigned from Anthropic on September 8, after also having worked at OpenAI, and explained his reasons in a long thread on X.
“None of the companies are acting responsibly,” he wrote. “They are running directly toward a self‑improving superintelligence.”
In the thread, Coxon argued that AI systems are approaching a point where they could “hack anything, revolutionize any field overnight, and acquire real power and resources”, and stated that the pace of progress in each of these domains “is not slowing down”.
He maintained that many of the people developing this technology privately fear that it could be fatal for humanity, even when they avoid saying so publicly.
“The people building AI systems sincerely believe these systems could kill us all by the end of the decade,” he wrote, adding that executives “soften how they say it in the media to sound reasonable”, while privately expressing the same fears.
Coxon’s departure is not an isolated warning.
Anthropic co‑founder Dario Amodei has spent much of the year issuing warnings about the very technology his company develops, cautioning in an essay published in January that a powerful AI could give authoritarian states tools for a “totalitarian nightmare” of complete surveillance, and later warning in a memo that superintelligence “will put us to the test as a species”.
OpenAI’s CEO, Sam Altman, has adopted a more moderate public tone, but continues to describe superintelligence — the term he prefers instead of AGI or artificial general intelligence — as something that will arrive “within a few thousand days”, meaning around 2032.
0 Comments