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

What the report says

A senior AI safety researcher at Anthropic has warned that there is a more than 10% chance artificial intelligence could kill all humans within the next decade, according to Sky News World. The report centers on a stark risk estimate from the company’s safety team rather than on a confirmed incident, and it frames the comment as part of wider debate over advanced AI systems and their potential dangers.

Anthropic is one of the leading firms working on large language models and other frontier AI tools. The researcher’s assessment highlights concerns often raised by safety experts about whether increasingly capable systems could become difficult to control, be misused, or cause large-scale harm if deployed without strong safeguards. Sky News World’s article, based on the available snippet, does not provide further technical detail about how the estimate was reached.

The warning matters because it comes from inside a major AI developer, adding weight to ongoing discussions about regulation, testing, and safety standards. More broadly, the claim reflects a growing split between AI’s commercial promise and the risks some researchers believe it may pose if development races ahead of protections. The limited source text does not indicate any immediate event or policy response.