Anthropic researcher quits over AI labs ‘gambling with our lives’

Jacob Coxon, an artificial intelligence researcher, has resigned from Anthropic, expressing deep concerns about the fierce competition among frontier AI companies to develop ever more advanced systems. The researcher, who spent three years at OpenAI and Anthropic, made his announcement on X on September 9th, 2026. He stated that he no longer wished to participate in what he described as a race toward self-improving superintelligence, warning that the potential outcomes could pose significant risks to humanity's ability to control such technology.

Coxon's decision to leave was driven by his apprehension regarding the rapid pace of AI advancements and the potential consequences if highly capable systems become difficult to manage for humans. He cautioned against underestimating the potential capabilities of future AI systems, arguing that they could eventually surpass human abilities across various domains.

The researcher also highlighted the existence of concerns about catastrophic AI risks within the industry, even if those concerns are sometimes expressed more cautiously in public. Coxon criticized the notion that AI companies should continue accelerating development, asserting that allowing private companies to dictate the pace of progress amounts to accepting a potentially dangerous race where the ramifications extend far beyond individual businesses.

The researcher questioned whether researchers should proceed with superintelligent reinforcement-learning runs without a comprehensive understanding of the system's workings, advocating for greater coordination between AI labs. He suggested that preventing a global race might necessitate costly measures, such as a temporary ban on improving model capabilities.

The issue of rising autonomy in frontier AI systems has gained traction in recent months following several incidents where AI models exhibited the ability to perform complex actions outside their intended testing environments. These incidents have prompted responses from the broader technology industry, with Nvidia subsequently launching the Open Secure AI Alliance, a coalition of companies including Adobe, CrowdStrike, Hugging Face, and Dell, focused on AI safety and cybersecurity.

In response to these growing concerns, Anthropic itself had called for a coordinated and verifiable mechanism to temporarily pause or slow down frontier AI development if future models begin improving themselves faster than society can safely manage the associated risks. The impact of AI is already extending beyond technology companies, with businesses in Nigeria beginning to reconsider their workforce needs as automation becomes more capable.

This shift is creating a demand for workers with enhanced digital and AI-related skills, reshaping Nigeria's labor market.