An Anthropic safety researcher says there is more than a 10% chance that advanced AI could lead to humans being killed, following the resignation of a colleague who raised safety concerns. The researcher characterizes the risk as potentially “killing all humans,” according to the report.
The outlets describe the same claim but with different emphasis. CNBC frames the comments around the colleague’s departure and focuses on the likelihood estimate and the researcher’s safety perspective. India Today presents the issue more directly as a “10% chance” scenario “next decade,” foregrounding the timing language used in the reporting. Neither outlet, in the provided material, offers independent verification of the probability figure or additional technical detail about how the estimate is derived.
Overall, the story centers on an internal safety-related dispute at Anthropic and the public discussion of existential risk estimates, but the specific basis, methodology, and broader consensus within the organization are not detailed in the excerpts.
