More than 10% chance AI 'could kill all humans' - Anthropic researcher

A senior researcher at Anthropic, a U.S. company developing the Large Language Model Claude, has warned that there is a more than 10% chance artificial intelligence could wipe out all humans within the next decade. Jacob Coxon, who previously worked at OpenAI, accused both OpenAI and Anthropic of racing to develop superintelligent AI without proper safety measures.

He argued that neither company is acting responsibly and that the technology could soon become superhuman, capable of hacking anything and acquiring vast resources. Coxon stated that staff at OpenAI, led by Sam Altman, have not fully grasped the civilizational risks, leading to continued rapid model improvements despite the dangers.

Anthropic's Alignment Science lead, Evan Hubinger, echoed Coxon's concerns, stating that Anthropic believes there is a 10% chance of AI causing human extinction by the end of the decade. Hubinger noted that Anthropic is working to prevent superintelligence but does not have a clear plan to achieve alignment. The Financial Times reported that Anthropic withheld its latest model from the UK's AI Safety Institute, while OpenAI, Anthropic, and Meta have faced hacking incidents involving their autonomous AI agents.

Despite these warnings, the U.S. government has not taken any steps to regulate AI development, with the Trump administration even supporting OpenAI in a lawsuit against the New York Times.