Jacob Coxon, a former Anthropic employee, argues that slowing AI development is futile due to alleged espionage and intellectual property theft.
Jacob Coxon walked away from his role at Anthropic last month with a stark warning: the industry’s high-stakes race toward superintelligence is built on a foundation that may already be compromised.
In an interview released Tuesday on the podcast hosted by Patrick Bet-David. Coxon—who left after accusing both Anthropic and OpenAI of gambling with human lives—argued that internal efforts to pace the development of advanced AI are largely ineffective. His reasoning centers on a claim of systematic espionage. asserting he is 90% certain that Chinese intelligence agents have already infiltrated the major U.S. labs.
“So if we continue, they’re going to have access to it,” Coxon said. “If we don’t continue, China’s going to create it themselves. So either way, China’s eventually going to have access to it.”
When pressed by Bet-David for specific identities or evidence. Coxon declined to name anyone. stating he was not interested in initiating a “witch hunt.” His allegations stand in direct opposition to the official stance at his former workplace. A source familiar with Anthropic stated that there is no evidence to support the claims. noting that the company maintains rigorous personnel vetting. strict access controls. and ongoing insider risk monitoring to mitigate foreign intelligence threats. Representatives for both Coxon and OpenAI did not respond to requests for comment.
Beyond human infiltration, Coxon pointed to technical maneuvers—specifically distillation attacks—as a primary driver for China’s AI advancements. This process. which involves using sophisticated models to train and refine less capable systems. has become a flashpoint between the global rivals. Earlier this summer, Anthropic formally accused Chinese labs, including Alibaba, Moonshot, DeepSeek, and Zhipu, of executing millions of such attacks.
These concerns surface as leadership at top labs. including Anthropic CEO Dario Amodei. OpenAI CEO Sam Altman. and xAI CEO Elon Musk. have publicly advocated for internal guardrails and a more measured pace to development. In a September post on X. Altman acknowledged that while progress remains rapid. the company intends to slow development slightly to prioritize safety. Critics. however. have questioned the timing of these calls for caution. suggesting they serve to build public anticipation for upcoming IPOs while creating barriers for smaller. emerging competitors.
For Coxon, the culture inside the labs remains a point of deep contention. He likened the workforce at Anthropic to a colony of worker ants, driven by an singular, intense focus on achieving superintelligence—a mission, he suggested, that blinds staff to the broader risks of the path they are on.
Anthropic OpenAI Jacob Coxon AI safety China artificial intelligence tech news