A researcher at Anthropic has resigned from the prominent AI safety company, warning publicly that some developers within the field believe artificial intelligence could ultimately kill all humans. The departure drew significant attention given that Anthropic was founded explicitly around the mission of building AI safely and has positioned itself as a more cautious alternative to other AI developers.

The researcher's exit underscores a deepening divide within the AI safety community itself — not merely between critics and boosters of the technology, but among those who have dedicated their careers to making it safer. The warning suggests that even inside organizations explicitly focused on risk mitigation, there is no consensus on whether current trajectories are acceptable.

Anthropic, co-founded by former OpenAI executives including Dario and Daniela Amodei, has attracted billions in investment and employs some of the most prominent figures in AI safety research. The company has publicly acknowledged what it calls "catastrophic" and "existential" risks in its own documentation, making the researcher's concerns consistent with language the company itself has used — though the resignation signals a belief that acknowledgment alone is insufficient.

The episode reflects a tension that has grown across the AI industry as capabilities have advanced rapidly: whether the companies best positioned to understand the risks are also the ones most likely to press forward regardless. Critics argue that competitive and commercial pressures make it structurally difficult for any lab to slow down unilaterally, even when internal voices raise alarms.

Anthropic did not immediately respond to requests for comment reported across multiple outlets. The resignation is likely to intensify scrutiny of AI safety practices at major labs and renew calls from some quarters for stronger government oversight of the sector.