-39.png)
Tech
Jacob Coxon, a researcher at Anthropic, resigned and warned that AI could kill all humans, saying he was concerned that AI companies were mishandling the risk. Evan Hubinger, another Anthropic researcher, backed the warning, stating he believes there is more than a 10% chance of this happening within the next decade.
The Guardian's short item frames the story similarly to the centrist outlets, emphasizing the extinction warning and the fact that a researcher resigned in protest of Anthropic and OpenAI 'ignoring or mishandling' the threat, with less elaboration due to article length.
Axios, Sky News, CNBC, and BBC report the core facts straightforwardly: Anthropic researcher Jacob Coxon resigned warning AI could kill all humans, and Evan Hubinger backed him up with a >10% p(doom) estimate. Axios and BBC add substantial context (p(doom) estimates from other experts, prior industry warnings, UK AISI model-withholding controversy, calls for treaties) framing this as part of a broader, escalating pattern of insider AI safety alarm that policymakers should heed, while still noting critics who suspect hype/marketing motives.
CoinDesk covers the resignation and warning in market/tech-industry terms; the Washington Examiner opinion piece pushes back, framing the story as overblown 'AI hysteria' and questioning the significance of a 10% probability claim.