Pakistan, Sept. 10 -- Anthropic insider Jacob Coxon has walked away from the AI industry with an accusation that deserves more than Silicon Valley's usual shrug. After working on training frontier models at both OpenAI and Anthropic, the 27-year-old researcher says the leading artificial-intelligence laboratories are racing toward self-improving superintelligence and "gambling with our lives."

While technologists have been predicting utopia and apocalypse in roughly equal measure for decades, it is disturbing to see where the latest warning comes from.

Coxon was not watching the AI boom from the outside. He was helping build it. Nor does he accuse Anthropic of being oblivious to the danger. Quite the opposite. His case is that researche...