Pentagon stops using Anthropic AI tools after blacklisting
The US Department of Defence has confirmed it stopped using Anthropic AI tools, months after designating the company a national security supply chain risk.
Stock photo for illustration only, not from the actual event
- The US Department of Defence has completely ceased using AI tools from Anthropic.
- Defence Secretary Pete Hegseth designated Anthropic a supply chain risk in February.
- Anthropic's Claude model was deeply embedded in Palantir's Maven Smart System.
- Anthropic has sued the Trump administration over the national security designation.
The US Department of Defence is no longer using artificial intelligence tools from Anthropic, an official has confirmed to the BBC, months after the company was branded a national security supply chain risk. A department official stated that the Pentagon has completely ceased the use of Anthropic products.
Defence Secretary Pete Hegseth initially labelled Anthropic a supply chain risk back in February, setting a deadline of late August to phase out its technology. However, multiple sources familiar with the matter, including former defence officials and contractors, told the BBC that the Claude model was still actively used for research, intelligence gathering, and military operations until recently.
The friction arose earlier this year when the Pentagon urged Anthropic to strip away safety guardrails from Claude to grant the military unfettered access. Anthropic refused, citing grave concerns over mass surveillance and autonomous weapons, which led Hegseth to slap the company with the national security supply chain risk designation typically reserved for foreign threat actors.
Stock photo for illustration only, not from the actual event
Anthropic slammed the move as unprecedented and unlawful, subsequently filing a lawsuit against the Trump administration to overturn the designation. As the legal battle unfolded in court, the Pentagon pivoted toward signing contracts with other tech giants including Google, xAI, and OpenAI, with OpenAI's tools seeing broader adoption across several military departments.
Removing advanced AI models from military architecture is a complex operational hurdle rather than a simple software toggle. Because Claude was deeply integrated into the Maven Smart System operated by Palantir, the prolonged timeline required to completely excise the technology highlights the deep reliance modern defence operations place on complex data platforms.
The Maven platform serves as the Pentagon's core apparatus for organizing satellite imagery and drone footage, feeding visual data directly into large language models to assist analysts in target identification and intelligence briefings. The fallout highlights the growing tension between national security demands and AI safety policies.
While the Pentagon has successfully severed ties with Anthropic and transitioned to alternative systems, the historic dispute establishes a profound precedent for how governments interact with commercial artificial intelligence developers over ethical boundaries and state security demands.
Source: BBC Business
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment