Secretary of War declares Anthropic a supply chain risk in viral tweet

Secretary of War Zvi Mowshowitz’s tweet labeling Anthropic as a “supply‑chain risk” has ignited a rapid policy scramble, with officials and industry insiders racing to assess the implications for the AI firm’s contracts with the Department of War (DoW). According to the original post on Thezvi, Mowshowitz’s declaration was accompanied by a lengthy “unhinged” statement that framed Anthropic’s involvement in DoW projects as a national‑security liability, citing concerns over “mass domestic surveillance” and the potential use of Anthropic models in lethal autonomous weapons. The post also referenced Dean Ball’s “Clawed” analysis as a prerequisite reading, underscoring the depth of the controversy (Thezvi).

The tweet prompted an immediate response from Anthropic’s leadership, which has historically emphasized robust supply‑chain safeguards. In a separate section of the same Thezvi article, the author notes that Sam Altman of OpenAI—Anthropic’s primary competitor—has been “excellent on the question of supply‑chain risk” but may need to do more, implying that Anthropic’s own risk‑mitigation practices are now under scrutiny. The piece contrasts OpenAI’s contract red‑lines, which reportedly “only forbid currently illegal activity,” with Anthropic’s “explicitly declined” terms that would have protected the company from certain DoW demands (Thezvi). This distinction suggests that Anthropic’s contractual posture may have left it more vulnerable to political pressure than OpenAI.

The political fallout was amplified by a brief but notable intervention from former President Donald Trump, who announced a six‑month wind‑down of Anthropic’s DoW contract. Reuters reported that Trump’s move was framed as a “de‑escalation” intended to give both the government and Anthropic time to transition away from the partnership (Reuters, Feb 25). The announcement, according to sources cited by Thezvi, was designed to avoid “extreme highly damaging options” that Secretary Hegseth and negotiator Emil Michael had reportedly floated. While the Trump‑era off‑ramp appears to have bought Anthropic a reprieve, it also signals a broader willingness within the administration to curtail AI firms deemed risky to critical infrastructure.

The broader context of the dispute reflects an escalating tug‑of‑war between the DoW’s demand for cutting‑edge AI capabilities and the industry’s pushback on ethical and security boundaries. Thezvi’s timeline outlines how the DoW had “many excellent options” before zeroing in on Anthropic, hinting at internal debates over whether to rely on commercial AI providers or develop in‑house solutions. The article also points to a “historically epic clusterfuck” that could have resulted had the DoW pressed forward without a compromise, underscoring the high stakes of integrating generative AI into defense systems.

Industry observers note that the episode may set a precedent for how future AI contracts are negotiated. Reuters’ coverage of the White House’s upcoming meeting with major AI firms—including Anthropic—on March 4 suggests that the administration is already seeking to formalize “red‑line” policies that balance innovation with security (Reuters, Feb 25). If Anthropic’s experience proves a cautionary tale, other vendors may push for more stringent contractual protections, mirroring OpenAI’s approach of limiting usage to “lawful” applications while retaining the right to object to certain government requests. The outcome could reshape the supply‑chain landscape for AI in defense, compelling firms to embed more robust risk‑assessment frameworks before entering federal agreements.

In the short term, Anthropic appears poised to survive the political storm. Thezvi’s analysis concludes that “Anthropic will probably survive,” citing the company’s resilience and the limited scope of the DoW’s reliance on its technology. Nonetheless, the episode has already heightened scrutiny of AI supply chains across the sector, prompting both regulators and vendors to re‑evaluate the balance between national‑security imperatives and the ethical constraints that underpin responsible AI deployment.