"All Lawful Purposes"
The Pentagon's demand that all AI models deployed for military use must be available for any legal application — the core contractual language at the heart of the dispute.
“All lawful purposes” was the key contractual language the Pentagon demanded from its AI providers. The phrase meant that any AI model deployed for military use should be available across the full spectrum of legal military operations — from logistics to intelligence to weapons development.
The Pentagon’s Argument
Under Secretary Emil Michael articulated the position clearly: “We have to be able to use any model for all lawful use cases.” The Pentagon argued that:
- Military ethical oversight should come from the Pentagon’s own established principles, not company policies
- Embedded restrictions in AI models could cause malfunctions during operations
- AI companies shouldn’t be making decisions about military strategy
The Problem
“All lawful purposes” is a maximalist demand. While it sounds reasonable — of course military tools should be usable for lawful purposes — the phrase elides critical questions:
- Who defines “lawful”? Military legality is distinct from civilian law
- Mass surveillance may be legally authorized but ethically problematic
- Autonomous weapons exist in a gray area of international law
- The phrase leaves no room for company ethics to differ from government policy
Industry Responses
Three of the four AI companies accepted the language. Only Anthropic maintained that certain lawful purposes crossed ethical red lines that the company would not cross regardless of legal authorization.