An Anthropic researcher has resigned from the artificial intelligence company, raising concerns about the rapid development of increasingly powerful AI systems and warning that the technology could pose serious risks if sufficient safeguards are not put in place.
Jacob Coxon, 27, said he left Anthropic after spending the past three years working in pretraining research at both Anthropic and OpenAI. In a post on X, he criticized the direction being taken by major AI companies as they compete to develop more advanced systems.
Coxon argued that Anthropic and OpenAI are moving too quickly toward what he described as self-improving superintelligence. According to his warning, the pace of development is not being matched by adequate safety measures.
He also made a stark claim about the concerns held by some people working in the AI industry. Coxon said that people building advanced AI systems genuinely believe the technology could potentially “kill us all by the end of the decade.”
His comments have added to a growing debate over the safety of advanced artificial intelligence. As AI models become more capable, researchers and technology experts have increasingly discussed how to ensure that future systems remain controllable and aligned with human interests.
Coxon accused the companies involved in the AI race of failing to act responsibly. He described their approach as “gambling with our lives” while competing to develop increasingly powerful artificial intelligence.
The researcher’s resignation is particularly notable because of his experience at two of the leading AI companies. His work in pretraining research gave him direct exposure to the development of large-scale AI systems and the processes used to make them more capable.
The debate surrounding AI safety has become more prominent as technology companies invest heavily in developing advanced models. Supporters argue that increasingly capable AI could deliver major benefits across areas such as science, education, productivity and research.
Critics, however, warn that systems capable of significant autonomous improvement could introduce risks that are difficult to predict or control. They argue that safety research and governance should progress alongside AI capabilities rather than after more powerful systems have already been developed.
Coxon’s remarks reflect the more cautious side of that debate. His central concern is that the race toward superintelligent AI could move faster than efforts to understand and control its potential consequences.
The resignation also highlights an internal challenge facing major AI companies: balancing rapid technological progress with concerns from researchers who believe stronger safeguards are necessary.
As companies continue developing increasingly sophisticated AI models, questions about safety, oversight and responsible development are likely to remain at the center of the technology debate.
Coxon’s warning does not establish that advanced AI will cause catastrophic harm, but it underscores the seriousness of concerns expressed by some researchers working directly in the field. The coming years could prove crucial in determining how the industry manages the risks associated with increasingly powerful artificial intelligence.
