TL;DR
The Pentagon has fully cut off Anthropic's Claude after months of continued use despite a February blacklisting, exposing how deeply AI tools embed themselves in military infrastructure and accelerating a shift toward OpenAI, Google, and xAI.
What happened
- Secretary of Defense Pete Hegseth designated Anthropic a "national security supply-chain risk" on February 27, 2026, setting a six-month phase-out deadline.
- The Pentagon confirmed to the BBC it "has ceased the use of Anthropic products" only after the BBC reported Claude was still actively in use as recently as last week.
- Claude was embedded in Palantir's Maven Smart System, the Pentagon's primary intelligence and data platform, where it processed satellite imagery, drone footage, and target identification tasks.
- Pentagon employees used Claude "throughout the whole period of the controversy," including during military operations against Iran, according to multiple former officials and contractors.
- The dispute originated when the Pentagon pressed Anthropic to remove safety guardrails from Claude; Anthropic refused, citing mass surveillance and autonomous weapons concerns, and has since sued the Trump administration.
Why it matters
- The months-long gap between the blacklisting order and actual removal proves AI integration is not plug-and-play: once a model is woven into a platform like Maven, extraction is slow and operationally risky.
- The Pentagon has now signed contracts with OpenAI, Google, and xAI as replacements, concentrating military AI dependency among vendors with fewer public safety constraints.
- Anthropic was the first advanced AI company to have tools deployed in classified government work, making its removal a significant reset of the US government AI vendor landscape.
- The blacklisting law used against Anthropic is "typically reserved for companies based in countries that pose a threat to the US," making its application to a US firm legally unprecedented and the subject of active litigation.
- Parallel to the Pentagon freeze, Dario Amodei met Trump twice at the White House last week, signaling a possible political thaw even as the legal fight continues.
What to watch next
- Anthropic's lawsuit against the Trump administration: a ruling overturning the supply-chain designation could reopen government contracts and set a precedent on AI safety refusals.
- OpenAI's expanding Pentagon footprint: two sources confirm broader military adoption in recent months, making its safety posture and willingness to remove guardrails the next pressure point to watch.
- Maven's transition: whether Palantir fully migrates to alternative LLMs without operational gaps will signal how replaceable any single AI vendor is in live military systems.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.