AI Could Threaten Humanity and Kill All Humans, OpenAI Researcher Claims in Resignation

Mian Nadeem

Published: September 09, 2026

Jacob Coxson, a 27-year-old researcher, resigned from leading AI companies OpenAI and Anthropic, warning of a dangerous global race to develop powerful, self-improving AI systems.

Coxson’s Concerns

Coxson, who spent three years researching AI training at these companies, argues that AI poses an existential threat to humanity, potentially leading to its extinction by the end of the decade. He cites a private fear among senior officials and researchers that the technology could spiral out of control.

According to a US magazine, Coxson claims that the race among AI developers is driven by the fear of lagging behind, even as they recognize the risks. He highlights the potential for future AI systems to surpass humans in various domains, including cyberattacks, and to gain rapid access to resources and power.

Safety Measures Urged

Coxson advocates for stricter safety measures in AI development, emphasizing the need to understand AI behavior before creating highly powerful systems. He suggests considering challenging steps like temporarily pausing capability improvements to avert a global race.

Endorsement from Anthropic

Evan Hubinger, head of alignment research at Anthropic, agrees with Coxson’s concerns. He acknowledges that Anthropic is working diligently on AI safety but admits they still lack a comprehensive solution, emphasizing the urgency of addressing the superintelligence alignment problem.

Expert Opinions

Despite Coxson’s resignation, experts caution that his concerns are not enough to halt the AI race. They highlight the growing awareness among technology companies of AI’s potential risks, yet the continued pursuit of advancement.

AI Could Threaten Humanity and Kill All Humans, OpenAI Researcher Claims in Resignation