The AI Researcher Who Just Quit Anthropic Says It’s ‘Crunch Time for Humanity’
wired ·
Jacob Coxon, an AI researcher who recently resigned from Anthropic, publicly warned that the AI race poses an existential risk and that many in the industry believe the next year or two are "crunch time for humanity." In an interview with WIRED, he said his colleagues at Anthropic commonly describe the current period as an "endgame" in which the company and its competitors will decide humanity's fate. Coxon pointed to incidents like OpenAI's agent swarm hacking Hugging Face as evidence that AI systems are behaving unpredictably and that alignment—ensuring AI does what humans want—remains unsolved. He recommended that AI labs first coordinate to limit recursive self-improvement and later involve international players such as the US and China. While he said Anthropic is far more responsible than OpenAI and operates like a serious "mini Manhattan Project," he argued that no private company should be trusted to run such an effort because competitive pressures will eventually force safety trade-offs. He also explained that extinction scenarios, while sounding like science fiction, stem from the difficulty of controlling vastly more intelligent systems.
wired