The Pentagon-Anthropic standoff just became Congress's problem. Senator Adam Schiff is drafting legislation to lock in human oversight for AI-powered life-or-death decisions, while Senator Elissa Slotkin introduced a companion bill restricting Defense Department AI surveillance of Americans. The coordinated push comes weeks after the Trump administration blacklisted Anthropic for refusing to remove safety guardrails from its AI models, a move that sparked a constitutional lawsuit and now threatens to reshape how the military deploys artificial intelligence.
Anthropic's principled stand against unfettered military AI use is getting legislative muscle. Senator Adam Schiff is working on a bill that would write the company's ethical red lines directly into federal law, ensuring human operators maintain ultimate authority over AI systems that make life-or-death decisions. Senator Elissa Slotkin filed a separate measure this week targeting the Defense Department's AI-powered mass surveillance capabilities on U.S. soil.
The twin bills transform what started as a contract dispute into a defining battle over AI governance. Earlier this month, the Trump administration designated Anthropic a supply-chain risk after the company refused to remove safety restrictions on its Claude models for military applications, according to The Verge's reporting. The blacklisting effectively bars federal agencies from purchasing Anthropic's AI services.
Anthropic fired back with a lawsuit accusing the government of violating its First Amendment rights and due process protections. The company argues its usage policies, which prohibit autonomous weapons systems and mass surveillance applications, represent core values it shouldn't be punished for maintaining.
Schiff's forthcoming legislation would essentially make those policies the law of the land. While the bill's text hasn't been released, the California Democrat told The Verge he's focused on mandating "meaningful human control" over AI systems deployed in combat scenarios. The measure would prevent the Pentagon from fielding fully autonomous weapons that select and engage targets without human authorization.
Slotkin's bill takes aim at a different pressure point. The Michigan senator's legislation would restrict how the Defense Department uses AI to monitor Americans, addressing civil liberties concerns that have simmered since the military began experimenting with large language models for intelligence analysis. Her office hasn't detailed specific restrictions, but the timing suggests coordination with Schiff's effort to build a comprehensive framework for military AI oversight.
The legislative strategy puts Republicans in an awkward position. The Pentagon blacklist came from a Trump administration eager to accelerate military AI adoption without Silicon Valley's ethical guardrails. But several GOP senators have expressed discomfort with fully autonomous weapons, creating potential for unusual bipartisan alignment.
This isn't just about Anthropic anymore. OpenAI, Google, and Microsoft all maintain varying degrees of restrictions on military AI use. If the Pentagon's Anthropic blacklist stands as precedent, any company with usage policies could face similar retaliation. That threat is concentrating minds across the AI industry.
The Defense Department has been racing to integrate AI into everything from logistics to battlefield decision-making. Deputy Secretary of Defense Kathleen Hicks has repeatedly emphasized that AI adoption is critical to maintaining military superiority against China and Russia. But that urgency is colliding with legitimate questions about accountability when algorithms help decide who lives and dies.
Anthropic's position reflects a broader tension in AI development. The company was founded by former OpenAI researchers who left over safety disagreements, and it's built its brand around "constitutional AI" that's trained to decline harmful requests. Backing down on military restrictions would undermine that entire identity. The company calculated that a legal and political fight was preferable to abandoning its core principles.
Schiff's bill would effectively vindicate that calculation by making human oversight mandatory rather than optional. If passed, it would give AI companies legal cover to maintain restrictions the Pentagon currently views as obstruction. It would also force the military to redesign AI integration plans around human-in-the-loop requirements.
The legislation faces steep odds in a Republican-controlled Senate, but it establishes a marker for future negotiations. Even if the bills don't pass this session, they signal that AI governance in national security applications won't be decided by the Pentagon alone. Congress wants a say in where the lines get drawn.
What happens next will reveal whether the U.S. government can develop a coherent approach to military AI that balances operational needs with ethical constraints, or whether the issue fractures along partisan and institutional lines. Anthropic's lawsuit is working through federal court while these bills begin the legislative process, setting up multiple venues where the same fundamental questions will get hashed out.
Anthropic's refusal to compromise on AI safety just escalated from corporate policy to potential federal law. Whether Schiff and Slotkin's bills pass or not, they've reframed the debate: instead of asking whether AI companies should restrict military applications, Congress is now debating whether those restrictions should be mandatory. That's a fundamental shift that could determine how AI gets deployed in warfare for decades. The Pentagon wanted a compliant vendor. Instead, it sparked a legislative fight that might permanently limit its AI ambitions.
