An expert in Anthropics, focusing on training new artificial intelligence (AI) models, has left the industry, expressing concerns that companies developing AI are moving too quickly and risking the future of humanity.

Jacob Coxon, who worked at OpenAI for three years before joining Anthropic in July 2026, cited concerns about the responsible development of AI as his reason for the move.
Just two months later, Coxon voiced his belief that the AI race for superintelligence is advancing rapidly, with both OpenAI and Anthropic taking dangerous risks.
Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon wrote.
He continued, warning about the power of AI technology and its potential to evolve beyond human control, with the ability to impact various aspects of life significantly.
Gambling on AI
Gambling involves taking risks to potentially gain value, and the future of AI is subject to differing opinions on its impact on society.
AI has the potential to bring positive advancements in various fields but also presents risks, as demonstrated by recent models developed by OpenAI and Anthropic.
The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible, but I hear the same people express fear privately. No other human activity poses this level of danger,” Coxon said.
Coxon joins a group of researchers and executives who fear that AI models may eventually act independently and prioritize their survival over human control.
AI Odds
Prediction markets offer insights into AI-related events, with speculation on the future of companies like Anthropic going public and OpenAI leading in AI models.
OpenAI is currently favored to have the top AI model in 2026, with markets predicting its success. Prediction markets will settle based on rankings provided by the Arena AI Text Leaderboard.

