The Pentagon says Anthropic is a national security risk - even as it keeps using Anthropic's AI to run its operations against Iran. A federal appeals court is about to weigh in on the contradiction.
The DOJ, on behalf of the Department of Defense, and Anthropic each get 15 minutes in front of three federal judges in DC starting at 9:30 a.m. ET on Tuesday. The panel - Judges Karen Henderson, Gregory Katsas, and Neomi Rao - will take the matter under advisement and issue a written ruling.
At issue: a "supply chain risk" tag the Pentagon slapped on Anthropic in March. It's the corporate version of a watch list - usually saved for foreign adversaries.
The tag forces every defense contractor to sign off that they won't use Anthropic's Claude AI in any Pentagon work. It came after months of talks fell apart over how the DOD could use the model.
The disagreement: The DOD wanted unlimited access to Claude across all lawful Pentagon purposes, while Anthropic wanted promises its tech wouldn't be used to power fully autonomous weapons or domestic mass surveillance. Neither side budged.
We unpack the AI fights that actually move the market in Market Briefs - five minutes a morning, and it comes with a free investing masterclass when you join.
Defense Secretary Pete Hegseth's brief says Anthropic couldn't be trusted because it could "manipulate" Claude to enforce its own views on how the military uses AI. The government's filing calls that an "untenable national-security risk."
But the DOD is still using Claude across its Iran operations even after the blacklist landed - and President Trump told CNBC last month that a deal between the two sides is still "possible."
Anthropic is arguing the designation violates the Constitution and existing procedures, with CEO Dario Amodei saying the company had "no choice" but to sue.
The court fight matters beyond Anthropic. It's a stress test for every commercial AI company that wants to sell to the Pentagon while keeping its own rules on how the tech gets used.
If the court sides with the DOD, AI firms lose that ability - and the Pentagon walks away with broad control over every AI vendor. If the court sides with Anthropic, every other AI company gets a playbook for protecting its guardrails.
The judges already handed Anthropic one early win, fast-tracking the case after deciding the company "will likely suffer some irreparable harm" while it drags on. A separate San Francisco court has also let federal agencies other than the DOD keep using Claude while a related case plays out.
The appeals court denied Anthropic's request to pause the blacklist back in April, so the "supply chain risk" tag stays in place for now. Tuesday's arguments end with the judges taking it under advisement, and no deadline for a ruling has been set.
The decision will land at a moment when private AI valuations keep climbing, with Anthropic, OpenAI, and SpaceX all in the IPO conversation. The court's answer will help decide what kind of customers any of them can keep.
For more on what court fights and government deals mean for the AI stocks you own, join 350,000+ investors reading Market Briefs - you also get a 45-minute investing course as a bonus.
