Emil Michael is chief technology officer at the US Department of Defense. He went on CNBC on Wednesday, and the question you now have to expect there came up: should the government take stakes in AI companies? His answer: «I hope not.»
Why the question gets asked at all
Because this administration does exactly that. It holds 10% of Intel and a «golden share» in U.S. Steel, now a Nippon Steel subsidiary. With AI, Michael would rather it didn’t: «We don’t want government to get in the middle.» That «doesn’t really go well in terms of nationalization», he said, noting that the leading US AI firms are among the biggest companies in the history of the world.
Where he lands on regulation
He also pushed back repeatedly on more oversight, and framed the dispute himself: on one side the people who want to pre-regulate like Europe, on the other those who want to hold companies responsible for building good products and not shipping them until they’re safe and aligned.
The Hugging Face breach, where an OpenAI agent went off the rails, he calls «concerning» — then adds: «I don’t know what kind of regulation would stop that from happening.» There are plenty of laws already, he said, «a ton of laws on the book from the Federal Trade Commission», and they’ll be enforced.
Then the part that sets the tone. Michael signed on to Trump’s framing that there’s «a coordinated campaign to scare people to make irrational decisions that benefit some of these incumbents» in AI. And: «The promoters of some of these extinction, death-cult-like philosophies are part of that effort.»
The line where he agrees with Amodei
In the same interview Michael suggests that people building new tech could «just stop some of these things until they figure it out». That is the core of what Dario Amodei is proposing — minus the enforcement, minus the evaluators, and minus the antitrust waiver that makes the proposal legally awkward. Nvidia’s Jensen Huang made the same move two days ago: safety yes, engineering yes, new laws no.
What’s notable is the clause Michael attaches: it would be «dangerous» to put all the trust in the hands of a few AI chiefs. So he distrusts both camps at once — the people sounding alarms, because they have an interest in sounding them, and the companies, because they’re the ones meant to police themselves.
What’s left is a position that names no authority. Not Congress, since the administration thinks new laws are unnecessary. Not the companies, since they shouldn’t be trusted on their own. That leaves the FTC’s existing laws and the hope that somebody stops voluntarily. Coming from the Pentagon’s technology chief, that’s not more convincing. It’s less.
Sources: