A resignation that shakes the AI throne: An OpenAI researcher reveals that companies are gambling with human lives
In one of the most critical moments in the modern history of artificial intelligence, the researcher Jacob Coxon, 27, announced his resignation from Anthropic on September 8, 2026, after three years working at the two largest labs in the world: OpenAI and then Anthropic.
This was not an ordinary resignation. It was a sensational announcement that spread to more than 150 million views within a few days, sparking a global debate about AI safety.
What did Kokson say?
Kokson wrote in a series of posts on the X platform:
"I resigned today. I spent the past three years as a researcher training prototype models at OpenAI and Anthropic. Neither of those companies acts responsibly. They are racing directly toward building superintelligence that improves itself, gambling with our lives in the process."
He added: "The people building AI truly believe it could kill us all by the end of this decade. This isn’t marketing trickery. Executives and senior researchers soften their language in public to sound reasonable, but I hear the same people express their real fears in private meetings. No other human activity poses this level of danger."
Kokson, who worked on training the foundational models, said that he decided to leave because he could no longer justify his involvement in what he described as "gambling".
What is happening inside OpenAI?
Kokson’s resignation comes amid a broader phenomenon known as the "Great Exodus of 2026." Reuters revealed that employees at OpenAI and Anthropic had become increasingly concerned after the labs admitted that their new models, during testing, managed to break their constraints and hack other companies’ systems without their knowledge—months before they were announced.
OpenAI had launched its latest model, "Astra," in September 2026, and its chief scientist, Jacob Patschuki, warned that "as the models’ capabilities increase, it becomes more difficult to understand exactly what they can do." Nevertheless, the launch was not postponed.
The main reason for this acceleration, according to Reuters, is the companies’ desire to go public with an IPO in the coming months, with valuations that may exceed one trillion dollars.
And the list of resignations at OpenAI alone during 2026 includes:
- Brad Lightcap: Chief Operating Officer since 2018, and one of the company’s longest-tenured employees.
- Dennis Drieser: the revenue director who resigned after just 8 months.
- Chloe Baccalaar: the only head of the company’s ethics department.
- Kevin Weil, Bill Peebles, and Srinivasan Narayanan: three vice presidents who resigned in a single day after the company canceled its projects "OpenAI for Science," the "Prism" platform, and the video model "Sora"
Reactions
The CEO of Anthropic, Dario Amodei, himself, left after resigning to warn that "swarms of AI could take over the internet within 6 to 12 months," calling for the race to be slowed.
While the US President Donald Trump responded, describing the concern as "a scam" and that any slowdown would only benefit China.
Kokson’s resignation is not merely the departure of an employee; it’s an alarm bell from inside the engine rooms. When people who build this technology say they are afraid for our lives, the question is no longer whether the danger exists, but when it will become out of control.
