A series of warnings from people linked to Anthropic says advanced AI could pose an existential risk. Multiple outlets report that a lead Anthropic researcher and other staff members have said they assess the chance of AI “killing all humans” within the next decade as more than 10%. The claims are tied to public posts and resignations by researchers who say major AI labs are not acting responsibly.

The outlets provide context around how the debate has shifted from whether AI is risky to how large that risk might be. Vox and Al Jazeera describe a broader fear among technologists and policymakers that rapid progress could create scenarios in which AI systems cause catastrophic harm. The Guardian and CBS News add detail on the statements attributed to Anthropic staff and explain that the warnings are largely about future development rather than specific, immediate mechanisms. The Guardian also frames the concern alongside reported resignations and names individuals involved. Slashdot and the other summaries emphasize that insiders point to the acceleration of capabilities and the possibility of systems gaining real influence over time.

While outlets vary in framing—ranging from prompting debate on governance to discussing potential scenarios—most sources describe the same core points: quantified risk estimates above 10%, warnings made publicly by Anthropic-connected researchers, and calls for urgent attention to AI safety.