OpenAI's Pentagon Contract Terms Allow 'Any Lawful Use' Including Potential Surveillance

✍️ OpenClawRadar📅 Published: March 2, 2026🔗 Source
OpenAI's Pentagon Contract Terms Allow 'Any Lawful Use' Including Potential Surveillance
Ad

OpenAI CEO Sam Altman announced that his company successfully negotiated new terms with the Pentagon, claiming the Department of Defense agreed to OpenAI's safety principles prohibiting domestic mass surveillance and requiring human responsibility for autonomous weapon systems. However, sources familiar with the negotiations say the Pentagon didn't actually budge on these issues.

The 'Any Lawful Use' Loophole

According to sources who spoke with The Verge, OpenAI's deal is significantly softer than what Anthropic was pushing for, largely due to three words in the contract: "any lawful use." The source confirmed that every aspect of OpenAI's terms boils down to: If it's technically legal, then the US military can use OpenAI's technology to carry it out.

Over the past decades, the US government has stretched the definition of "technically legal" to cover sweeping mass surveillance programs. The Pentagon reportedly wouldn't back down on its desire to collect and analyze bulk data on Americans during negotiations.

Ad

Anthropic's Stance and Consequences

Anthropic was blacklisted by the Department of Defense for standing firm on two specific red lines:

  • No mass surveillance of Americans
  • No lethal autonomous weapons (AI systems with the power to kill targets without human oversight)

Defense Secretary Pete Hegseth designated Anthropic as a supply chain risk for this refusal. OpenAI's former head of policy research, Miles Brundage, commented on X that "in light of what external lawyers and the Pentagon are saying, OpenAI employees' default assumption here should unfortunately be that OpenAI caved + framed it as not caving, and screwed Anthropic while framing it as helping them."

Contract Details and Industry Reaction

Altman used the Trump Administration's preferred name for the Defense Department, calling it the Department of War (DoW) in his statement. He claimed that "The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement."

Across social media and the AI industry, people immediately challenged Altman's claim, questioning why the Pentagon would suddenly agree to these red lines when it had previously stated it would never do so. OpenAI spokesperson Kate Waters denied that the agreement allowed for crossing certain lines, stating that "The system cannot be used to..." though the statement was cut off in the source material.

📖 Read the full source: HN AI Agents

Ad

👀 See Also

Bohrium AI Proteomics Competition 2026 with $13K Prize and Compute Support
News

Bohrium AI Proteomics Competition 2026 with $13K Prize and Compute Support

Bohrium is running an AI proteomics competition in 2026 with a $13,000 prize pool, internship opportunities, and compute support. The competition was discussed on Hacker News with 17 points and 5 comments.

OpenClawRadar
AWS Lambda MicroVMs: VM-level isolation for user and AI-generated code, with suspend/resume up to 8 hours
News

AWS Lambda MicroVMs: VM-level isolation for user and AI-generated code, with suspend/resume up to 8 hours

AWS launches Lambda MicroVMs, a serverless compute primitive built on Firecracker, providing per-user VM isolation, near-instant launch, and up to 8-hour state preservation for executing user/AI-generated code.

OpenClawRadar
Claude Code v2.1.83 adds managed settings fragments, transcript search, and security improvements
News

Claude Code v2.1.83 adds managed settings fragments, transcript search, and security improvements

Claude Code v2.1.83 introduces a managed-settings.d/ directory for team policy fragments, transcript search with / and n/N navigation, and CLAUDE_CODE_SUBPROCESS_ENV_SCRUB=1 to strip credentials from subprocess environments. The release also includes CwdChanged/FileChanged hooks, sandbox.failIfUnavailable setting, and fixes for macOS exit hangs, UI freezes, and memory leaks.

OpenClawRadar
Reddit Discussion on Long-Term Risks of Coding Agent Dependency
News

Reddit Discussion on Long-Term Risks of Coding Agent Dependency

A Reddit user argues that current coding agents like Claude Code and Copilot create dependency that could lead to vendor lock-in, centralization of software creation, and commoditization of engineering craftsmanship.

OpenClawRadar