Trump’s 60 Minutes interview touched on AI safety, threat to arrest an AI exec

On 60 Minutes, President Trump said the dangers AI executives keep warning about are smaller than they claim and that the risk is already being handled. His example of handling it was striking: he said the Justice Department called the head of one of the big AI companies, after a report of something bad being done with AI, and told that person to get ready for jail. He would not say who, or what the conduct was.

If that account is accurate, the most powerful lever over frontier AI in the US right now may be private, unexplained pressure from the top, not published rules. He also said he would work with Democrats on guardrails, but not at the cost of slipping behind China. Those three ideas sitting together tell builders, investors, and the public a lot about how the next few years of AI policy may actually be run.

I want to separate two things here. The first is the substance of the risk debate. Reasonable people disagree about how near or severe the worst AI failure modes are, and a leader saying the danger is manageable is a defensible position on its own. We tend to lean toward letting useful systems ship and fixing problems as they appear.

The second thing is the method, and that is where I get uneasy. Managing risk by phoning an executive and threatening prison, with no named company, no stated offense, and no public process, is the opposite of the predictable oversight that lets good AI get built. The President says the person changed everything afterward. We have no way to check what changed, whether it made anything safer, or whether the original report was even true.

Scale that up and the trajectory is clear. Labs learn that the real compliance department is political proximity. Companies in favor get latitude; companies out of favor get calls. Safety becomes whatever the person on the other end of the phone wants that week, and the public sees none of it. That is bad for safety advocates and bad for builders who just want stable rules.

The China framing deserves a note too. The line that China has no guardrails except whatever Xi says, followed by a remark that his own control is strong but not quite that strong, frames centralized authority as the benchmark. I would rather the US compete on transparent rules that markets and courts can see. The openness to bipartisan guardrails is the most hopeful line in the interview, and I'm watching to see whether it turns into anything written down.

The comments came in a 60 Minutes interview in which Ross Douthat asked about AI executives' public warnings over their own models. Trump also repeatedly said AI despite earlier dismissing the term, and floated calling it SI, for super intelligence, when talking about China's Xi Jinping.