Trump AI memo and OpenAI stake plan reshape US governance strategy

The Trump administration is simultaneously stepping on the gas and tinkering with the steering wheel of the AI race. The new national security memo codifies what was already happening informally: frontier models like GPT‑5.5‑Cyber and Claude Mythos are being pulled deep into intelligence and defense workflows, with an explicit directive to accelerate their use in targeting, planning and analysis — so long as they aren’t deployed for domestic mass surveillance or censorship. That’s a political constraint, not a technical one, and history suggests such red lines can move over time.

Layered on top is a far more radical idea: the federal government taking equity stakes in leading AI labs, potentially via a sovereign wealth fund seeded with OpenAI shares. If implemented, this would blur the line between regulator and shareholder in an industry already struggling with concentration and conflicts of interest. It could also give Washington direct leverage over an AGI developer’s governance at precisely the moment its systems become most powerful. For the race to AGI, this combination of deep national‑security integration and partial state ownership would make US frontier labs even harder to dislodge globally — but it would also fuse their incentives with the geopolitical aims of a single government to an unprecedented degree.