Anthropic researcher resigns, warns self-improving AI risks extinction

reported by 2 outlets· 2 articles · confidence: high · first seen 2026-09-09 15:02 UTC

What this means for you

Nothing to act on yet. One researcher's warning and resignation is not a product change. If you build or deploy self-improving systems, his call for pacing agreements may become a policy demand.

Anthropic researcher Jacob Coxon has resigned, saying self-improving AI systems pose an extinction risk and calling for pacing agreements between AI labs. In an interview with Wired, he describes a “mini Manhattan project” inside Anthropic and says labs have only a few years left to make their systems safe.

Key facts

  • ·Jacob Coxon resigned from Anthropic. source
  • ·He cited AI extinction fears as his reason for resigning. source
  • ·He called for pacing agreements between AI labs. source
  • ·He described a “mini Manhattan project” inside Anthropic. source
  • ·He said AI labs have only a few years left to make their systems safe. source

What the sources say

  • TechCrunch AI (press)Reports the resignation and his call for pacing agreements between labs.
  • Wired AI (press)Adds interview detail on his internal project description and safety timeline warning.

Sources

The original reporting. Follow these — they did the work.

Related stories

← the wire