A researcher at Anthropic is reported to believe there is a greater than 10% chance that artificial intelligence could kill all humans, in the latest warning about the potential safety risks posed by the technology.
The claim adds to a series of increasingly serious concerns being raised about artificial intelligence and its possible long-term dangers. The supplied material does not identify the researcher, provide the full remarks, or set out the evidence or reasoning behind the estimate.
Artificial intelligence safety has become a prominent subject of public debate as researchers, companies and policymakers assess the possible consequences of more powerful AI systems. In this case, the warning is framed in stark terms, with the researcher said to regard a catastrophic outcome as more than a remote possibility.
No further details are provided about when or where the comments were made, whether they were part of an interview, a public statement or a research discussion, or how Anthropic has responded.
The latest known position from the supplied material is that the warning stands as part of a broader pattern of rising concern over AI safety, but no additional official reaction or proposed next steps are included.