Anthropic backs DoW mission, fights designation

This is less a routine PR cleanup than a hard reset of Anthropic’s public posture on defense work. The company is trying to reassure customers, calm Washington, and avoid letting OpenAI define the Pentagon AI narrative alone.

• –Anthropic is drawing a bright line between objection to the designation itself and support for defense use cases like intelligence analysis, cyber operations, and planning
• –The statement narrows the practical blast radius by arguing most customers are unaffected unless Claude is used directly inside Department of War contracts
• –Amodei’s apology over the leaked memo signals Anthropic wants to de-escalate the feud without backing off its two policy red lines on autonomous weapons and mass surveillance
• –For AI developers and contractors, the key takeaway is that model access risk now sits much closer to geopolitics, procurement rules, and agency politics than ordinary product reliability
• –The backdrop of the Pentagon’s OpenAI deal makes this look like a strategic platform fight over who becomes default AI infrastructure inside the US defense stack