Anthropic CEO Dario Amodei issued a statement on Monday addressing a recent designation by the Department of War that labels the artificial intelligence company as a supply chain risk to U.S. national security. The company received the formal notice on March 4 and plans to contest the decision through legal action, describing it as unfounded.
Amodei emphasized that the designation has a limited reach, impacting only the direct integration of Anthropic’s Claude AI model into contracts with the Department of War. He noted that the overwhelming majority of customers, including those with other federal ties, face no restrictions on using the technology. This interpretation aligns with the wording in the department’s letter and stems from the narrow provisions of the governing law, 10 U.S.C. 3252, which aims to safeguard government operations using the minimal necessary restrictions.
The statute, Amodei explained, focuses on protecting public interests rather than penalizing providers, and it explicitly bars broader limitations on AI usage or partnerships unrelated to specific military contracts. For more details on the legal framework, see 10 U.S.C. 3252.
Prior to the announcement, Anthropic had engaged in constructive discussions with the Department of War about potential collaboration paths that respect the company’s strict policies against fully autonomous weapons and widespread domestic surveillance. Amodei highlighted the firm’s pride in its contributions to military efforts, including tools for intelligence assessment, simulation exercises, strategic planning and cybersecurity operations that aid troops in active engagements.
The CEO clarified that Anthropic has no intention of influencing tactical choices, leaving those to military professionals. He reiterated positions outlined in recent company updates, such as the statement from Thursday and the response on Friday, underscoring the AI firm’s commitment to ethical boundaries in high-stakes applications.
Amodei also expressed regret over an internal company message that surfaced in media reports the previous day. He stressed that Anthropic did not authorize its release and that the document, drafted amid a flurry of developments including presidential and departmental announcements plus a Pentagon partnership with rival OpenAI, captured raw frustration rather than the leadership’s measured perspective. Written six days earlier, it no longer represents the current dynamics.
Moving forward, the priority remains supporting national defense personnel during critical missions. Anthropic pledged to supply its models to the Department of War and security agencies at a symbolic price, backed by technical assistance, until a full handover can occur or as long as legally feasible.
Despite the rift, Amodei affirmed strong alignment with the department’s goals of bolstering American security through AI innovation across federal agencies, vowing that all upcoming steps will build on this mutual dedication.
