Anthropic Clashes With Pentagon Demands
- Buster Wurm
- 2 days ago
- 1 min read
Anthropic is marketing its AI models to the U.S. national security community but faces a key obstacle: the Pentagon demands broad freedom to use these tools, while Anthropic insists on strict, hard-coded limits.
The dispute centers on Claude Gov, a customized version of Anthropic’s Claude, built for government clients to process classified materials, intelligence, and cybersecurity data. Negotiations to expand Pentagon work have stalled amid a clear divide: Anthropic seeks explicit technical limits preventing use for mass surveillance or autonomous weapons without human oversight, while the Pentagon wants broad operational flexibility, emphasizing legal compliance. The conflict is whether legal standards alone suffice or if technical restrictions are also needed.
Tensions increased when Claude, through Anthropic’s partnership with Palantir, was reportedly used in a U.S. military operation in Venezuela targeting former President Nicolás Maduro and his wife. Airstrikes hit sites in Caracas, according to people familiar with the matter. Anthropic’s guidelines ban violence, weapons development, and surveillance. The company declined to comment on specific operations, but insisted any use must follow its policies.
Pentagon spokesman Sean Parnell said the Defense Department’s relationship with Anthropic is under review. He framed the decision as to whether partners will support warfighters in future conflicts. The dispute now threatens a $200 million contract. Officials are considering canceling it amid wider debate over AI regulation. Anthropic CEO Dario Amodei has called for strict guardrails, especially against lethal autonomy and surveillance, which can clash with military goals as AI’s role grows.
Sources:
Comments