What happened: The Guardian reports Anthropic sued the US Department of Defense after it was blacklisted from government work, arguing the decision violated its First Amendment rights and escalated a months-long standoff.

Why it matters: This is a rare, explicit collision between AI safety “red lines” and defense procurement. If vendors must accept “any lawful use,” then the practical meaning of voluntary safety commitments gets stress-tested in court, not blog posts.

Wider context: The piece argues Silicon Valley’s posture toward military contracts has shifted, with lucrative deals and political realignment changing what companies say they will or won’t support — and forcing competitors to choose sides.

Background: Anthropic has sought to block its models being used for domestic mass surveillance or fully autonomous lethal weapons. The Guardian says the Pentagon’s demands would permit broad “any lawful use,” which Anthropic frames as incompatible with its safety principles.

Anthropic-Pentagon battle shows how big tech has reversed course on AI and war — The Guardian

Singularity Soup Take: Big Tech’s “we have values” phase is entering its “please see Exhibit A” phase. The real question isn’t whether AI will be used for war — it’s which companies get paid, and what disclaimers survive contact with procurement language.

Key Takeaways:

• Lawsuit escalation: The Guardian says Anthropic sued the DoD after being blacklisted, claiming retaliation tied to its stance on how its AI models can be used and framing the dispute as a constitutional issue.
• Safety constraints: Anthropic is described as trying to prohibit uses like domestic mass surveillance or fully autonomous lethal weapons, while resisting demands that it allow “any lawful use” of its technology.
• Industry shift: The article argues that past anti-military postures in tech have softened as defense contracts grow more lucrative, making Anthropic’s refusal a test case for whether “guardrails” are principles or marketing.