A Defense Department official told the BBC on 5 October 2026, in a statement, that the Pentagon “has ceased the use of Anthropic products.” The BBC reports that multiple people familiar with the matter, including former US defence officials and contractors, said Anthropic’s Claude model was still used by the department as recently as last week in research, analysis and intelligence gathering and in military operations against Iran; the BBC says multiple sources also said it was embedded in Palantir’s Maven Smart System. The BBC says an Anthropic spokesman declined to comment on the statement.
What the Department Said
The BBC reports that a Defense Department official said in a statement that the Pentagon “has ceased the use of Anthropic products.” The BBC says the statement came on Monday, after multiple people said in interviews that Claude was still being widely used in recent weeks. It also says the details of the department stopping all use of Anthropic’s tools had not been previously reported.
The BBC reports that Defense Secretary Pete Hegseth designated Anthropic a “national security supply-chain risk” in February and announced that the Pentagon would stop using it by late August. This publication read the statement only as the BBC quotes it.

Business Pill · SOMETHING HIDDEN INSIDE SOMETHING ELSE
A one-minute explainer of the idea behind this story: an embedded dependency. It teaches the concept, not this story’s figures.
The key insight: The department’s statement, as the BBC quotes it, is about ceasing the use of Anthropic products, while the BBC’s sources describe Claude as used as recently as last week and as embedded in Palantir’s Maven. Lauren Kahn, as the BBC quotes her, says tools that become integrated “can be painful to remove.”
What Was Posted on X
The Information’s Leo Schwartz summarises the interview he and Amir Efrati did with Pentagon official Emil Michael. Per his post, he told them the administration has a range of options, including export controls, if AI companies do not cooperate, and that the Defense Department will continue to bar Anthropic as long as it has a “moral principle” embedded in its software that allows its models to exert control over humans. The update above quotes the lines in his post.
What the Sources Told the BBC
The BBC reports that multiple people familiar with the matter told it that as recently as last week Claude was still used by the department in research, analysis and intelligence gathering, as well as in military operations against Iran. It says those people include former US defence officials and contractors who have worked closely with the Pentagon on AI, and that several spoke on condition of anonymity because they were not authorised to speak to the media. The BBC says Claude was viewed as a critical tool for sorting through vast quantities of information, the people said.
The BBC says multiple sources told it that Claude was embedded in Maven Smart System, a data platform operated by Palantir that the BBC describes as the Pentagon’s primary platform for organising intelligence and other data. It reports that sources familiar with the way the technology is used said analysts and other officials rely on Maven to access and organise visual data, including satellite imagery and drone footage, and that the data was then fed into Claude and other large language models for tasks such as identifying potential military targets.
According to the BBC, one source who was given a demonstration of Maven by Palantir earlier this year said officials frequently used the technology when compiling one-page briefs.
One of the people told the BBC that Pentagon employees continued to use Claude tools, including Anthropic’s Mythos models, “throughout the whole period of the controversy.” Another person familiar with Maven told the BBC that recent private demonstrations of Maven for government officials have also included Claude. The BBC adds that Cameron Stanley, who leads AI work within the department, said during a demonstration of Maven earlier this year that it is deployed “across the entire department.”
What a Department Official Said in September
DefenseScoop reported on 11 September that Emil Michael, the Under Secretary of Defense for Research and Engineering, told a small group of reporters that the Pentagon was close to moving its classified systems off Anthropic models. DefenseScoop quotes him as saying: “I’d say about 90% has transitioned. All of the Maven Smart Systems or Palantir work has been transitioned months ago, and we’re on track to get it done by the end of the month.”
DefenseScoop also quotes Michael as saying that one of the mistakes was that the Pentagon “was solely reliant on one model,” and that “the number one idea is to get multiple models.” It reports that the department uses commercial frontier models in both classified systems and unclassified tools, with suppliers, deployment methods and security levels differing by workspace.
The two accounts use different words. Michael, as DefenseScoop quotes him, spoke of the Maven and Palantir work having been transitioned months earlier. The BBC’s sources spoke of Claude being used as recently as last week, and of its being embedded in Maven. This publication has not seen a document that reconciles them.
The Background the BBC Gives
The BBC reports that Anthropic had been used by the US government and military since 2024 and was the first advanced AI company to have its tools deployed in government agencies doing classified work. It says that earlier this year the Pentagon pressed Anthropic to remove its safety guardrails from Claude and give the military unfettered access, and that Anthropic refused, citing concerns about mass surveillance and autonomous weapons. The BBC says Anthropic called the designation “unprecedented and unlawful” and has sued the Trump administration to try to overturn it.
According to the BBC, Hegseth set a six-month deadline to phase out Anthropic on 27 February, one day before the US and Israel attacked Iran. The BBC says that as the legal fight has played out in the courts, the Pentagon has signed contracts with other major tech companies, including Google, xAI and OpenAI, and that two people familiar with AI use in the department said OpenAI tools have been more widely adopted in recent months in some military departments.
A former senior US defence official told the BBC that when Hegseth first ordered the phase-out, “there was concern with moving beyond Anthropic to OpenAI” and other tools.
An Analyst on Integration
The BBC quotes Lauren Kahn, a senior research analyst at Georgetown’s Center for Security and Emerging Technology and a former US defence official, as saying that because Claude was deeply integrated into Maven it would have been difficult to remove it quickly from the system. “The fact that it took the Pentagon until now to remove all use of Claude from across the system is indicative that these things are not just plug and play,” she said, adding: “Once they become integrated it can be painful to remove them.”
Kahn’s comments are an analyst’s view of how integration works, as the BBC quotes them. The general cost of leaving a tool once it is built into a workflow is the subject of this publication’s Business Pill on switching costs.
Update, 19 October: Pentagon’s Michael Threatens Export Controls, The Information Reports
Emil Michael told The Information in an interview published on 19 October that he would use a range of executive tools, including export controls, to check the behavior of AI companies if they do not cooperate with the government, according to the publication. The Information reports that in the same interview Michael compared frontier labs that call for more regulation to drunk drivers who ask someone else to pull them over.
The Information describes Michael as the Trump administration official who led the public feud with Anthropic, pushing for the Defense Department to label the company a supply chain risk after it refused to remove safety guardrails. It reports that he is now one of the four senior officials leading the Super Intelligence Force that Trump announced this month, and that he is charged with acting as the government’s liaison to Anthropic and other AI companies on protocols for cyber breaches, hacks and other critical incidents. The interview, with Amir Efrati and Leo Schwartz, is also on The Information’s YouTube channel, whose description says it covers the government’s clash with Anthropic and how the government could enforce AI safety rules.
As we read it, this moves the dispute described above from one department’s supplier choice toward tools that could reach any lab, though The Information’s summary does not say which companies or which export-control measures Michael had in mind.
In a post on X, The Information’s Leo Schwartz, who conducted the interview with Amir Efrati, listed more of what Michael told them. He wrote that his public feud with Anthropic will not get in the way of his new duties as Super Intelligence Task Force vice chair, where he is tasked with working with AI companies on an incident reporting framework (“If that means I’m talking to someone who’s suing me, well then I’m talking to them,” as quoted by The Information’s Leo Schwartz on X). He also wrote that the Defense Department will continue to bar Anthropic as long as it has a “moral principle” embedded in its software that allows its models to exert control over humans, and that he said frontier labs are acting like “drunk drivers” who demand someone else pull them over when they call for stricter government regulation.
Correction, 19 October: an earlier version of this update attributed this X post to The Information’s Stephanie Palazzolo. It was posted by Leo Schwartz, who conducted the interview with Amir Efrati. We have since written up the full interview: Pentagon Floats Export Curbs on AI Labs, Keeps Anthropic Out.
The Structural Read
The BBC report rests on a statement from a Defense Department official and on accounts from several people familiar with the matter, some of whom, the BBC says, spoke on condition of anonymity. It also quotes Lauren Kahn, a named analyst, and Cameron Stanley, who leads AI work within the department.
The BBC reports that multiple sources said Claude was embedded in Maven, a platform operated by Palantir. DefenseScoop quotes Emil Michael as saying in September that all of the Maven Smart Systems or Palantir work had been transitioned months earlier. This publication has no document that shows how the two accounts line up.
The BBC reports that the Pentagon announced it would stop using Anthropic by late August, and that the statement to the BBC came on Monday 5 October. Kahn, as the BBC quotes her, says integrated tools are not plug and play.
BBC, quoting Lauren Kahn of Georgetown’s Center for Security and Emerging Technology
“Once they become integrated it can be painful to remove them.”
Three Implications
WHAT THE STATEMENT COVERS The BBC quotes the official as saying the Pentagon “has ceased the use of Anthropic products.” The reports this publication read do not say which systems or dates the statement covers, and this publication cannot say either.
WHERE THE SOURCES PLACE CLAUDE The BBC’s sources describe Claude as used in research, analysis and intelligence gathering and, according to several of them, inside Palantir’s Maven. These are the sources’ accounts as the BBC reports them.
WHAT REMAINS UNKNOWN No department document, contract text or Anthropic comment is in the reports this publication read, and the BBC says an Anthropic spokesman declined to comment. None of it has been independently checked.
What Is Not Established
This publication has no document from the Defense Department, no contract text and no comment from Anthropic: the BBC says an Anthropic spokesman declined to comment when asked about the statement. It read the department’s statement only as the BBC quotes it.
According to the BBC, several of its sources spoke on condition of anonymity, and this publication cannot test their accounts. It also cannot say whether the department’s statement, the September remarks and the sources’ accounts describe the same systems or the same dates.
This publication read the BBC report of 5 October 2026 and the DefenseScoop report of 11 September 2026. It did not contact the Defense Department, Palantir or Anthropic.
The Bottom Line
A Defense Department official told the BBC that the Pentagon “has ceased the use of Anthropic products.” The BBC reports that multiple people familiar with the matter said Claude was still used as recently as last week and that it was embedded in Palantir’s Maven. The BBC says an Anthropic spokesman declined to comment. This publication verified none of it independently and has no document that settles between the two accounts.
91,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. We are drawing on a BBC report from October 2026 and a DefenseScoop report from September 2026, both of which we read. We encountered the Defense Department’s statement only as the BBC quotes it, and we did not contact the Defense Department, Palantir, or Anthropic. For the 19 October update, The Information’s article is paywalled and we did not read it in full: we relied on the publication’s free summary and the title and description of its interview video, and Michael’s remarks are given as The Information reports them. The few words in quotation marks are as quoted by The Information’s Leo Schwartz in his 19 October post on X. We have not checked any of these claims independently. Nothing we have written is a forecast, and none of it should be read as investment advice.
Sources: DefenseScoop (1) · DefenseScoop (2) · The Information: Pentagon’s Emil Michael Vows Export Controls if AI Labs Don’t Cooperate With Government (19 Oct 2026) · The Information on YouTube: Pentagon CTO on Anthropic, AI Safety and ‘Super Intelligence’ (19 Oct 2026)









