Anthropic Researcher Jacob Coxon Quits AI Over Self-Improvement Race Fears

On September 9, two resignations by frontier-lab researchers drew attention. According to the Wall Street Journal, Anthropic researcher Coxon left the AI industry out of concern that his own lab and its competitors were racing to build systems that could spiral out of control and pose existential risks to humanity, stating clearly that he did not want to take part in the industry-wide sprint toward "self-improving AI systems." Former OpenAI policy chief Miles Brundage shared the report, sparking debate about the tension between frontier labs' safety commitments and their commercial incentives. The same day, another researcher, Jacob Hilbert Spaess (online handle hilbertspaess), announced his resignation from Anthropic. He said he had spent the past three years doing pretraining research at OpenAI and then Anthropic, and his core allegation is that both companies are "acting irresponsibly," charging full speed toward self-improving superintelligence and effectively "gambling with our lives." Confirmed

• WSJ reported that Anthropic researcher Coxon left the AI industry over concerns about the race toward out-of-control AGI
• Jacob Hilbert Spaess publicly announced his resignation from Anthropic and stated he had done three years of pretraining research across OpenAI and Anthropic
• Spaess publicly accused both frontier labs of irresponsibly racing toward self-improving superintelligence

Why it matters

• The departing researchers came from Anthropic, widely seen as "safety-first," and its direct competitor OpenAI; a string of insider exits erodes confidence in frontier labs' ability to police themselves
• Both individuals share the same core concern: the commercial race is overriding safety considerations around out-of-control self-improving systems, extending the safety controversy from outside critics to frontline researchers themselves

2026-09-09 ~ 2026-09-09 · 9 related posts

• WSJ: Anthropic researcher quits AI industry over fears of uncontrollable AGI race — peterwildeford · 2026-09-09
• [source] Researcher quits Anthropic after 3 years in pretraining at OpenAI and Anthropic, blasting both labs — peterwildeford · 2026-09-09
• [source] Exec quits over industrywide rush to build self-improving AI, citing humanity-ending risk — Miles_Brundage · 2026-09-09
• [source] Anthropic researcher Jacob Coxon quits AI over fears self-improving models could go uncontrollable by 2027 — rohanpaul_ai · 2026-09-09
• Anthropic researcher Jacob Coxon quits AI over fears self-improving models could go uncontrollable by 2027 — rohanpaul_ai · 2026-09-09
• Anthropic Researcher Quits Over AI Fears, WSJ Reports — Bubbly-Air7302 · 2026-09-09
• Anthropic researcher Jacob Coxon resigns, warning superintelligence could kill us all by 2030 — Polymarket · 2026-09-09

2 near-duplicate retellings: Miles_Brundage · peterwildeford