I just came across a pretty crazy story about how Trump is handling the AI company Anthropic, and honestly, it sounds like a joke if it weren’t for the real consequences.

Last Friday, the government banned Anthropic from continuing to work with the federal government, and also prohibited all contractors from doing business with them. That’s an extreme penalty usually reserved for hostile nations. The reason? Anthropic refused to let the Department of Defense use their AI system Claude for citizen surveillance or developing autonomous weapons that can make decisions to eliminate humans.

Secretary of Defense Pete Hegseth is behind this request. Just so you know, he’s the same person who messaged the planning group to punish astronauts for speaking correctly, wanted to punish pilots for speaking the law, and even shot toy buses with powerful lasers despite FAA warnings that this could blind pilots flying with passengers. Yet, this same person wants unrestricted access to advanced AI technology.

When Anthropic refused, they were blacklisted. And here’s the really concerning part: the Trump administration is building a public surveillance system similar to China’s. Trump says he wants to catch up with China through “global technological dominance” and “widespread use of AI.” China has built one of the most advanced public surveillance systems in the world, collecting facial recognition and biometric data from citizens against their will. Now, Peter Thiel’s Palantir—who supports Trump—is developing a similar platform for US law enforcement.

But wait, there’s an important detail here. Anthropic signed a $200 million contract with the Department of Defense, which clearly prohibits using Claude for mass surveillance or developing autonomous weapons. Hegseth is demanding these restrictions be lifted. But Anthropic refused. CEO Dario Amodei said the company cannot “with a clear conscience” agree to that.

Then, Deputy Secretary of Defense Emil Michael publicly claimed that Anthropic “lied,” because the DoD “does not do mass surveillance since that is already illegal.” But that sounds like a joke when you consider the history of this regime.

I’m not an AI expert, but if Hegseth controls Claude, this technology could be used to synthesize billions of messages, emails, and posts simultaneously, create digital intelligence profiles on anyone, automatically track where people go, and support large-scale law enforcement. That’s exactly what Trump is trying to build.

The dishonesty of this regime doesn’t help. If Trump can change contract terms after the fact, why sign the contract at all? I suspect we’ll see major legal disputes from Silicon Valley over this issue.