A 27-year-old AI researcher who spent three years working on frontier models at OpenAI and Anthropic just walked away from one of the most coveted jobs in tech, and he did it loudly. Jacob Coxon resigned from Anthropic on September 8, 2026, publishing a thread on X that accused both companies of treating humanity’s safety like an acceptable trade-off in the race to build smarter machines.
Five days later, Coxon appeared on NBC News’ “Meet the Press” to make his case to a broader audience: AI labs need to coordinate globally, and they need to do it before the window closes.
The case for alarm
Coxon’s departure wasn’t a quiet two-weeks-notice situation. His public criticism centered on what he described as a gamble with humanity’s safety, warning that catastrophic outcomes from advanced AI could materialize by the end of the decade. In his telling, both OpenAI and Anthropic have prioritized speed of development over the kind of thorough safety assessments that the technology demands.
What makes the critique harder to dismiss is who else is saying it. Evan Hubinger, Anthropic’s alignment science lead, has acknowledged the company lacks a clear strategy for solving alignment problems related to superintelligence. Hubinger has publicly estimated a greater than 10% probability that AI could lead to the eradication of humanity within the next decade.
