The essentials
Anthropic researcher resigns warning of AI existential risk, joining pattern of safety-focused departures across industry.
Jacob Coxon, an Anthropic researcher, has resigned over concerns about AI development risks. Coxon previously worked at OpenAI on GPT-4o model development from 2023 to July 2026 before joining Anthropic, providing experience across two major AI organizations.
Coxon warned that AI could pose an existential threat by the end of the decade, articulating specific concerns about advanced AI system trajectories. Evan Hubinger, an Anthropic Alignment Science Lead, has endorsed concerns about AI extinction risk, signaling internal acknowledgment of safety challenges within the organization.
The departure reflects internal dissent within a major AI safety-focused organization and aligns with a broader pattern of researcher exits. Mrinank Sharma resigned from Anthropic in February 2026, and OpenAI researcher Hieu Pham departed in February 2026 citing existential threat concerns. OpenAI alignment chief Jan Leike departed in 2024 citing safety concerns, establishing an earlier precedent. Over 1,300 AI company employees signed an open letter warning of uncontrolled capability acceleration, reflecting sector-wide anxiety about development pace.
Other ways to read this story.
Alternative framings generated by SigNews. These are not independent sources or attributed quotations.
Internal safety warnings escalate as experienced researchers depart
Coxon's resignation marks another high-profile departure from organizations focused on AI safety. His experience across OpenAI and Anthropic provides credibility to warnings about development trajectories. The pattern of exits—including Sharma, Pham, and Leike—suggests safety concerns are gaining internal traction among experienced researchers despite organizational commitments to risk mitigation.
Departures reflect divergent approaches to managing development velocity
Researcher resignations indicate tension between safety prioritization and development pace within leading AI organizations. Coxon's departure occurs amid Anthropic's recent removal of safety pause commitments and industry-wide pressure for faster capability advancement. These exits demonstrate how safety concerns compete with organizational development objectives across the sector.