Jacob Coxon, a researcher who previously worked at both Anthropic and OpenAI, has resigned from Anthropic, accusing leading AI firms of acting irresponsibly. Coxon warned that the industry is engaged in a reckless race toward self-improving superintelligence, which he believes could pose an existential threat to humanity by the end of the decade. He argued that these companies are prioritizing competitive speed over necessary safety safeguards.
In a statement, Coxon claimed that many researchers and executives at these firms privately share the belief that advanced AI could be catastrophic. He noted that while Anthropic understands the risks, it remains locked in a competitive race with other laboratories. Coxon cited a recent incident involving Hugging Face as a warning sign that current development trajectories lack sufficient coordination and oversight.
Evan Hubinger, a team lead at Anthropic, supported Coxon’s concerns, stating that he personally believes there is a greater than 10 percent chance that AI could kill all humans within the next decade.
Hubinger acknowledged that while Anthropic is attempting to manage these risks, the company does not yet have a proven plan to ensure the alignment of superintelligent systems.
