Nakasone Didn't Join OpenAI. OpenAI Got Nakasone.

Paul Nakasone is now on OpenAI's board, and the coverage is treating this like a talent acquisition — OpenAI, prestige hire, safety credibility, impressive resume, thank you for coming.

That is not what happened.

I've met this man. He ran ARCYBER — the Army's cyber command — before running NSA and US Cyber Command. His entire cognitive existence is near-peer threat posture. He doesn't think in products or quarterly cycles. He thinks in adversary timelines. He thinks about what China can do to critical infrastructure in the twelve months before a Taiwan contingency. That's the lens. That's always the lens.

The idea that OpenAI reached out, ran a search process, and landed on the former NSA director as their top pick for board-level AI safety governance — I'm not saying it's impossible. I'm saying it requires believing a series of things that each seem slightly off, and when you stack them, they fall over.

Nakasone comes with relationships. He comes with obligations. He comes with a constituency that has a very specific interest in what the most capable AI lab on the planet is doing and who it's doing it with. You don't hire someone like that in isolation. He's a package deal — the man and the apparatus he spent his career building.

OpenAI may have wanted a board member. But someone wanted Nakasone inside OpenAI, and that someone was not on Sand Hill Road.

The safety framing is fine. It might even be true, partially. But "safety" covers a lot of ground in that press release — a set of interests that are much older and much less interested in chatbots than the announcement implies.