Jacob Coxon, fresh off his exit from Anthropic, has decided that the best way to warn humanity about existential risk is through the BBC. The message: AI researchers are genuinely frightened that current development speeds could end humanity. The market response: Series D rounds are oversubscribed.
This is the core absurdity that nobody in the funding ecosystem wants to acknowledge. A researcher credible enough to work at one of the field’s most serious safety-focused labs is now on record saying the thing might actually kill everyone. Investors read this and saw a buying opportunity.
Why would they care about extinction-level risk when the next eighteen months of hockey-stick growth projections are so compelling? The venture thesis is simple: if AI development continues at current speeds and causes civilizational collapse, at least the cap table will have been optimized.
Coxon’s argument hinges on the rate of development being the problem—not AI itself, but the velocity. This is the kind of nuance that dies in a Series C pitch meeting. “We’re moving too fast and it could destroy humanity” translates to investors as “the market is moving fast and we need to move faster.” The fear becomes a feature, not a bug.
The real terror isn’t that AI researchers are frightened. It’s that their fear has become completely decoupled from how the industry actually operates. A man says the world might end. Three venture firms immediately call his old colleagues asking what the Series E round looks like.