When an AI lab’s ethics rules become a Pentagon law fight

The U.S. Senate is translating Anthropic’s corporate AI ethics into federal law. Sen. Adam Schiff (D-CA) is drafting legislation to "codify" the company’s red lines on autonomous weapons and mass surveillance, ensuring human oversight in life-and-death decisions. The bill emerges from Anthropic’s public conflict with the Pentagon, which escalated when the Trump administration blacklisted the company for refusing to compromise its ethical guidelines.

Schiff’s proposal aligns with Sen. Elissa Slotkin’s (D-MI) AI Guardrails Act, which explicitly limits the Defense Department’s ability to use AI for nuclear detonations or tracking U.S. citizens without human approval. Both bills reflect growing congressional unease about unchecked military AI, particularly after high-profile disputes over Anthropic’s refusal to deploy its models for certain Pentagon applications. The legislation targets what Schiff called "illicit purposes"—a category that includes fully autonomous weapons systems and domestic surveillance tools that operate without human judgment.

The source material also shows that the bills arrive as AI’s role in national security faces unprecedented scrutiny. Slotkin’s AI Guardrails Act would prohibit the Defense Department from using AI to detonate nuclear weapons or conduct domestic surveillance without human intervention, addressing two of the most contentious applications of military AI. The measures also respond to public sentiment, with Schiff noting that Anthropic’s stance reflects "policies that the vast majority of American people agree with."

Anthropic’s legal battle with the government adds urgency to the legislative push. After the Trump administration designated the company a supply-chain risk, Anthropic filed suit alleging constitutional violations. The case underscores the tension between corporate AI ethics and military demands, a conflict now playing out in Congress. If passed, the bills could establish a framework for regulating AI in national security, potentially influencing how other nations approach military AI governance.

The legislation’s scope remains fluid, with details still under negotiation. However, its core premise—human control over AI in critical decisions—mirrors Anthropic’s internal policies and could become a benchmark for future federal AI regulations. Observers are watching whether the bills will gain bipartisan support or face opposition from defense hawks advocating for unconstrained AI development.

For source context, compare The Verge AI, NIST AI RMF and FTC AI guidance.