Anthropic Researcher Resigns Warning Against Dangers of Self-Improving Artificial Intelligence

9 Sept 2026, 8:32 pm · 3d ago · 1 min read · TechCrunch AI

Jacob Coxon, a research scientist at artificial intelligence safety lab Anthropic, has resigned from his position over acute concerns regarding potential AI-driven existential risks. Coxon publicly warned against the uncontrolled acceleration of recursive, self-improving artificial intelligence models without robust safety frameworks in place. In his resignation statement, he strongly urged leading frontier AI research laboratories to establish formal pacing agreements and safety pause protocols. His departure underscores growing internal tensions among researchers at top AI companies who fear commercial competition is overshadowing critical safety verifications and alignment safeguards before advancing toward superhuman intelligence.