So here's what happened on Friday in the increasingly public fight over who gets to build AI for the U.S. military, and under what rules. In one corner, you have the Pentagon reportedly shaking hands with OpenAI (a key Microsoft partner) on a set of safety terms. In the other, you have President Donald Trump ordering the federal government to stop using technology from rival Anthropic (backed by Amazon). It's a split decision that immediately turned into a political and industry shouting match.

The Pentagon's Reported Deal with OpenAI

According to reports, the Pentagon has agreed in principle to follow the safeguards OpenAI proposed for using its artificial intelligence models in classified environments. Now, to be clear, no contract is signed yet. But the framework they're talking about is pretty specific.

OpenAI's rules reportedly say: you can't use our AI for mass surveillance. You can't use it for autonomous weapons. The models have to stay in secure cloud environments—think heavily guarded server farms—and can't be baked directly into the hardware of things like fighter jets or drones (what the industry calls "edge systems"). The company also wants its own security-cleared researchers to have a seat at the table for continuous monitoring and to advise on risks.

The idea, as CEO Sam Altman reportedly told employees, is to put strong guardrails in place and then learn and adapt from real-world use. It's a "move fast with safety harnesses" approach for the most sensitive applications imaginable.

Trump's Order to Phase Out Anthropic

This news broke just hours after a move from the other end of Pennsylvania Avenue. President Trump directed all federal agencies to stop using Anthropic's technology. The stated reason? He called the company's restrictions on lawful military applications "unacceptable." Agencies currently using Anthropic's products, like its Claude assistant, have six months to transition away.

Defense Secretary Pete Hegseth backed the move, arguing the military must have "full, unrestricted access" to AI systems for lawful defense. He went a step further, announcing plans to officially designate Anthropic as a supply-chain risk to national security. That's a formal label that can severely limit a company's ability to do business with the government.

The core of the conflict is Anthropic's own ethical red lines, which are publicly known and are strikingly similar to the ones OpenAI is reportedly pushing with the Pentagon. Anthropic has also drawn a hard line against the use of its AI for mass surveillance or in autonomous weapons systems.

Silicon Valley and Wall Street Weigh In

When the government picks a favorite in a tech rivalry, people tend to have opinions. This was no exception.

Ilya Sutskever, the co-founder and former chief scientist of OpenAI, took to social media to call it "extremely good" that Anthropic had not backed down from its principles. He added that it was "significant" OpenAI had taken a similar stance, framing it as a win for responsible AI development across the board.

Over on Wall Street, SkyBridge Capital founder Anthony Scaramucci saw it differently. He described the episode as the government "bullying" a private company for setting its own terms of service.

Investor Ross Gerber, known for his outspoken views, took a more direct commercial stance. He said he directed employees at his firm to use Anthropic's Claude model "as much as possible," a small but symbolic act of support.

The reaction wasn't confined to finance and tech. Senator Elizabeth Warren (D-Mass.) demanded transparency, posting that the American public "deserves to know" what Pentagon officials are planning. She called on Defense Secretary Hegseth to testify about the decision, signaling this debate is headed for congressional hearings.

So, where does this leave us? With a Pentagon potentially partnering with an AI giant under a specific set of rules, a White House banning a competitor for having similar rules, and a whole lot of very powerful people arguing about what it all means for national security, corporate autonomy, and the future of a technology that's still being defined. The battle lines for military AI aren't just being drawn in boardrooms or classified briefings anymore; they're being litigated in real-time on social media and in press releases. Buckle up.