Anthropic finalizes classified NSA contract for secret AI surveillance tools

Anthropic, the company behind the Claude AI models, is putting the finishing touches on a classified contract with the National Security Agency (NSA) to build secret surveillance tools. The deal, first reported by the New York Times, was cleared by the White House over explicit objections from the Pentagon, highlighting internal government friction over the use of frontier AI in intelligence operations.

The contract comes at a time when spy agencies are grappling with an AI chip shortage, making partnerships with leading AI labs even more critical. Neither Anthropic nor the NSA have disclosed the scope of the tools, but experts expect the system to leverage Anthropic’s safety-focused Claude models to process and analyze massive datasets — likely including signals intelligence. The Pentagon had raised concerns about ceding too much AI capability to a single contractor, but the White House sided with the NSA on urgency. This deal marks one of the most direct integrations of a major AI company into core surveillance infrastructure.