Pentagon warns of national security risks from Anthropic's foreign workers
The US Pentagon raised national security concerns about Anthropic's employment of foreign workers, including Chinese nationals, in filings related to Anthropic's lawsuit against the government. The conflict stems from disagreements over military use of Anthropic's Claude AI system and led to threats of contract loss and a federal ban on using Anthropic technology.
What happened
In court documents filed as part of Anthropic's lawsuit against the US government, Pentagon Deputy Secretary Emil Michael stated that Anthropic's employment of foreign nationals creates 'national security risks.' Michael stated that Anthropic employs numerous foreign citizens to develop and support its large language model products, and noted that among these workers are Chinese nationals.
Michael asserted that foreign worker employment increases 'hostile risk levels' if such workers comply with China's National Intelligence Law. The Pentagon also stated that risks from other major US AI companies that employ foreign workers are reduced through 'responsible and reliable conduct' consistent with technical and security assurances provided when working with the Pentagon, but characterized Anthropic's situation as different.
Reported impact
The U.S. military has reportedly used Claude for intelligence activities, weapons development, and military operations. The dispute between the Pentagon and Anthropic arose after the company refused to make Claude available for 'all legal military purposes' and raised security concerns about military use.
Index analysis
The Pentagon's concerns center on Anthropic's employment of foreign nationals, particularly Chinese citizens, in roles developing large language models that the military relies upon.
Governance controls
The Secretary of Defense issued an ultimatum on February 24 demanding Pentagon access to Anthropic's AI systems by February 27 or face loss of government contracts. President Donald Trump ordered all federal agencies to immediately cease use of Anthropic technology on February 27.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.