How a Math Superstar Who Fears AI Ended Up at OpenAI

A person who genuinely believes AI could end humanity just accepted a job at the company most responsible for pushing that frontier. If that doesn't make you pause, I don't know what will.

I came across this story about a famous mathematician—someone whose entire life is built around rigorous logic and proof—who has spoken openly about existential AI risk. And yet, they're now working at OpenAI. Not as an outside consultant, not as a critic on some advisory board, but as an employee. That contradiction is exactly why I can't stop thinking about it.

Why would someone walk into the fire?

The obvious interpretation is the "if you can't beat them, join them" angle. But I think it's more nuanced than that. There are a few reasons that make sense to me:

• Seat at the table: If you're genuinely scared of AI, the worst place to be is outside the lab yelling into the void. Being inside means you see the actual research, the safety discussions, the unspoken trade-offs before they become public crises.
• Access to compute and data: Alignment research on the outside often suffers from lack of resources. Inside a frontier lab, you get to test real-world safety hypotheses on models that actually matter.
• Maybe they think the fears are overblown? That would be the simplest reading, but the article suggests otherwise. They're supposedly