Pentagon Formalizes 'Any Lawful Use' Contracts with Seven AI Companies for Autonomous Military Systems

The U.S. Department of Defense announced Friday that seven AI companies have signed contracts permitting the military to deploy their technology for “any lawful use” in classified environments, according to The Guardian. The companies are SpaceX, OpenAI, Google, Nvidia, Reflection AI, Microsoft, and Amazon Web Services. Anthropic, which rejected the clause over concerns about autonomous lethal weapons and domestic surveillance, remains excluded.

This follows NCT’s earlier reporting on the Pentagon’s AI vendor realignment after Anthropic’s supply-chain risk designation in March.

What Changed

The April 29 story covered the Pentagon expanding classified AI access through individual vendor negotiations. Friday’s announcement formalizes the arrangement: all seven companies have now agreed to identical “any lawful use” contractual language, and the Defense Department confirmed their models will be integrated into “Impact Levels 6 and 7” network environments for classified operations.

The Pentagon stated that these agreements “accelerate the transformation toward establishing the United States military as an AI-first fighting force and will strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” according to The Guardian.

The $54 Billion Autonomous Weapons Budget

The contracts land alongside the Pentagon’s $54 billion budget request for autonomous weapons development, reported The Guardian. The “any lawful use” language removes vendor-imposed restrictions on how these models can be applied within military operations, including autonomous targeting, drone warfare, intelligence synthesis, and decision support across classified and unclassified networks.

Anthropic’s Position

Anthropic rejected the “any lawful use” clause over concerns its technology could enable fully autonomous lethal weapons or domestic mass surveillance. The Pentagon designated Anthropic a supply-chain risk in March, the first time an American company has received that label, and barred contractors from using its products. Anthropic sued in response.

Defense Department officials believe the seven-company alliance could pressure Anthropic back to the negotiating table, The New York Times reported. That leverage is complicated by Anthropic’s Mythos model, whose autonomous vulnerability-detection capabilities have rattled government officials assessing classified cybersecurity postures.

Reflection AI’s Inclusion

One notable signatory is Reflection AI, a two-year-old company that has not yet released a publicly available model. The company is building open-source models positioned as a counter to Chinese AI firms like DeepSeek and is seeking a $25 billion valuation, The Guardian reported. Its backers include Nvidia and 1789 Capital, the venture fund where Donald Trump Jr. is a partner.

Autonomous Agents in Classified Defense

The formalization of “any lawful use” contracts means these seven vendors’ models can now power autonomous agents operating within the Pentagon’s most restricted classified networks without vendor-imposed guardrails on use case scope. For the autonomous systems industry, this establishes the U.S. military as the largest single customer for unrestricted AI agent deployment, with budgetary authority exceeding the entire commercial agentic AI venture market.