White House Eyes Executive Order Against Anthropic as OpenAI Robotics Chief Quits Over Pentagon Deal

Caitlin Kalinowski, OpenAI's head of robotics and consumer hardware, announced her resignation on March 7 via a post on X. She had spent 16 months building the company's physical AI program since joining from Meta in late 2024. Her departure came just days after OpenAI secured a controversial agreement to deploy its AI models within the Pentagon's classified cloud networks. Kalinowski's resignation post was blunt: "This wasn't an easy call," she wrote, explaining that OpenAI's contract failed to establish adequate guardrails against two specific dangers she considers existential. First, the potential for AI systems to conduct warrantless mass surveillance of American citizens. Second, the risk of granting AI systems "lethal autonomy" without human oversight.

The OpenAI contract emerged from a failed negotiation between the Department of Defense and Anthropic. According to TechCrunch reporting, the Pentagon had designated Anthropic as a "supply-chain risk" after talks collapsed over military control of AI models, particularly regarding autonomous weapons and domestic surveillance applications. When Anthropic's $200 million deal fell apart, the DoD pivoted to OpenAI instead. OpenAI's agreement, revealed earlier in the week, involves deploying the company's AI models on classified Pentagon systems. While specific use cases remain classified, the scope appears to include both surveillance capabilities and potential integration with weapons systems. This represents a significant shift for a company that once explicitly banned military applications in its usage policies.

On March 9, Anthropic filed a federal lawsuit against the Department of Defense, challenging the agency's decision to label it a supply-chain risk. The complaint, filed in the Northern District of California, argues the designation was arbitrary and politically motivated. Anthropic's legal team contends the Pentagon retaliated after the startup refused to grant military control over its AI models. The lawsuit has already drawn support from dozens of current and former OpenAI employees, as well as engineers from Google DeepMind and Microsoft.

The Trump administration is preparing an executive order targeting Anthropic, according to White House sources who spoke to Wired. The potential order, still in draft form, could impose additional restrictions on the startup's government contracts and data access. Press Secretary Karoline Leavitt declined to rule out "further action" against Anthropic when asked directly at Tuesday's briefing. The move comes just two days after Anthropic filed its federal lawsuit, suggesting the administration intends to escalate rather than negotiate. Legal experts say such an order could complicate Anthropic's court challenge by creating new regulatory hurdles. The White House has not indicated when the executive order might be signed.