People keep asking what the Pentagon is doing with Claude, and with Claude Code, right now. Here is the real-time answer as of Friday, November 20, 2026, based on what the Pentagon, the courts and the press have actually said.
The Short Answer: Officially, Nothing
In early October 2026 a Defense official told the BBC that the department “has ceased the use of Anthropic products.” That statement ended a six-month phase-out ordered back in February. So, by the Pentagon’s own account, there are no current activities at the U.S. military that run on Claude.
And on Claude Code specifically, the tool developers use to write and run software with Claude: there has been no public reporting of the Pentagon using it at all. Anyone claiming to list “everything the Pentagon is doing with Claude Code today” is guessing.
Two caveats are worth knowing. The Pentagon has never given an exact date when the last Claude system was switched off. And, according to BBC sources, Claude was still running inside Palantir’s Maven system only about a week before the Pentagon made that statement.
What Claude Did Inside the Military (Before the Ban)
- Classified networks: in July 2025 Anthropic and the Pentagon signed a contract that made Claude the first frontier AI model approved for the military’s classified networks, under Anthropic’s usage policy.
- Maven Smart System: Claude was built deeply into Maven, the Pentagon’s main platform for organising intelligence and other data, which is run by Palantir. That integration is why removing it took months.
- Operations: Claude was reportedly used during the U.S. operation in Venezuela in early 2026, and BBC sources said it was still being used in operations against Iran until shortly before the October statement.
In other words, Claude’s military role was mainly helping analysts sort and summarise huge amounts of information, not flying drones or pulling triggers. Those are exactly the uses Anthropic had refused to allow.
Why the Pentagon Dropped Claude
The fight was not about hacking or foreign ownership. It was about two lines Anthropic would not cross:
- No mass surveillance of Americans at home.
- No fully autonomous weapons.
The Pentagon wanted to use Claude for “all lawful purposes” with no company-imposed limits. When Anthropic let the February 27, 2026 deadline pass, President Trump ordered federal agencies to phase out Anthropic’s technology, and Defense Secretary Pete Hegseth labelled the company a national security supply chain risk, the first time that label had ever been used against an American company. There is a full breakdown in this explainer on why the Pentagon blacklisted Claude and what the ban covers.
Where the Court Fight Stands
- August 2026: a California judge struck down one of the designations, the one issued under a defense procurement law. That ruling still blocks the administration’s broader restrictions.
- September 25, 2026: the U.S. Court of Appeals for the D.C. Circuit upheld the other designation, 2 to 1, under the 2018 supply chain law. Judge Henderson dissented.
- Next: Anthropic says it disagrees with the ruling and could ask the Supreme Court to take the case. As of today we have not found confirmation that it has filed.
What Replaced Claude
Palantir’s chief technology officer, Shyam Sankar, says Claude has “already been replaced” with no operational impact. The Pentagon now uses models from xAI, OpenAI and Google, plus open-weight models such as Nvidia’s Nemotron.
Meanwhile, Anthropic went public on Nasdaq this week under the ticker ANT. You can follow the stock live in my ANT live chart and RSI post. For more on the company behind Claude, see this overview of Anthropic and how Claude is built and a rundown of what Anthropic’s listing means for investors. And if you have ever wondered what it costs to stand up the kind of computing power these models run on, there is a breakdown of what an AI data center really costs and the AI Build Configurator for pricing your own build.
The Bottom Line
Right now the Pentagon says it is using Claude for nothing, and there is no public sign it ever used Claude Code. What Claude did do, while it was allowed, was help the military’s analysts make sense of intelligence on classified systems. It lost that job because its maker refused to let it be used for domestic mass surveillance or for weapons that pick their own targets. Whatever you think of that choice, that is the real story.
Frequently Asked Questions
Is the Pentagon still using Claude?
No, according to the Pentagon itself. In early October 2026 a Defense official told the BBC the department “has ceased the use of Anthropic products”, after a six-month phase-out ordered in February.
Does the Pentagon use Claude Code?
There is no public reporting of the Pentagon using Claude Code specifically. The reported military uses of Claude were through Anthropic’s classified-network contract and Palantir’s Maven Smart System, and the department says all Anthropic use has now stopped.
Why did the Pentagon ban Claude?
Anthropic refused to drop two limits on Claude, no mass domestic surveillance of Americans and no fully autonomous weapons, while the Pentagon demanded use for “all lawful purposes”. Defense Secretary Pete Hegseth then labelled Anthropic a national security supply chain risk.
What AI does the Pentagon use instead of Claude?
Palantir’s CTO said Claude has been replaced by models from xAI, OpenAI and Google, plus open-weight models such as Nvidia’s Nemotron.
Sources: BBC (Pentagon statement, October 2026), CNBC and ABC News (D.C. Circuit ruling, September 25, 2026), The Next Web (Palantir CTO interview). Status checked November 20, 2026.
The cover graphic is original work by Kevin Katovic. © 2026 Kevin Katovic. All rights reserved.