Anthropic, OpenAI clash over Pentagon AI safeguards

This is less a PR spat than a power struggle over who sets military AI norms.

• –Anthropic’s Feb 26 statement frames its stance as pro-defense but against mass domestic surveillance and fully autonomous weapons with current model reliability.
• –OpenAI’s Feb 28 post says its contract includes legal and policy constraints, then added clarifying language on March 2 around domestic surveillance of U.S. persons.
• –The conflict shows how “lawful use” language can still produce major interpretation gaps between labs and government buyers.
• –For developers, this signals more pressure for auditable safeguards, policy controls, and human-in-the-loop enforcement in defense-adjacent AI deployments.