WASHINGTON, D.C., : The U.S. Department of Defense (DoD) is weighing a formal “supply chain risk” designation against the San Francisco-based artificial intelligence company Anthropic following a dispute over the military use of its flagship AI model, Claude. The disagreement centers on Anthropic’s refusal to remove certain safety guardrails that limit how its technology can be deployed in defense operations.
Defense Secretary Pete Hegseth is reported to be close to issuing the designation, a move that would significantly affect Anthropic’s standing within the U.S. defense contracting ecosystem.
Background of the Dispute
The disagreement follows a classified U.S. military operation conducted in January 2026 involving former Venezuelan President Nicolás Maduro. According to individuals familiar with the matter, the U.S. military used Anthropic’s Claude model during the planning and execution stages of the mission.
Access to Claude was not obtained directly from Anthropic. Instead, the system was integrated into military workflows through Palantir Technologies, a major defense data contractor that had pre-existing integration pathways with the AI model.
The operation reportedly involved lethal kinetic actions, including bombings in Caracas that resulted in casualties. Following the mission, an Anthropic executive contacted a counterpart at Palantir to inquire whether Claude had been used in ways that contributed to lethal outcomes. The inquiry reflected concerns that such usage could conflict with Anthropic’s published terms of service.
Anthropic has publicly denied discussing operational specifics of the Venezuela mission with the U.S. government. However, the company’s usage policies explicitly prohibit the use of Claude to facilitate violence or conduct surveillance activities.
Policy Differences Over Military Use
After the January operation, the Pentagon approached major AI providers—including Anthropic, OpenAI, Google, and xAI—seeking assurances that their systems could be employed for “all lawful purposes” in military contexts.
Defense officials have argued that this standard is necessary to ensure operational flexibility across a wide range of missions. According to officials, restrictions embedded in commercial AI models may create ambiguity in classified or fast-moving environments.
Anthropic has declined to adopt the “all lawful purposes” framework. Chief Executive Officer Dario Amodei has maintained two non-negotiable guardrails for Claude’s deployment:
-
The model cannot be used in the development or deployment of fully autonomous lethal weapon systems that operate without meaningful human oversight.
-
The model cannot be used for domestic mass surveillance, including processing open-source intelligence or social media data for large-scale monitoring of American citizens.
In addition to these restrictions, Anthropic has refused to remove usage limits because of its stated corporate governance framework and safety commitments. The company has consistently emphasized that its AI systems are designed to align with defined safety principles intended to reduce the risk of misuse in high-impact scenarios. Company leadership has indicated that removing guardrails for military clients would create a precedent that could weaken enforcement of its broader safety standards across all deployments.
Anthropic’s position is also based on its internal policy that AI systems should not be directly involved in lethal decision-making processes or large-scale surveillance infrastructures. According to individuals familiar with the company’s stance, leadership believes maintaining consistent safeguards across both commercial and government clients is necessary to uphold contractual terms, investor commitments, and its publicly stated responsible AI framework.
Defense officials contend that these constraints introduce operational “gray zones” and limit the military’s ability to integrate AI tools seamlessly into defense planning and intelligence analysis systems.
Potential ‘Supply Chain Risk’ Designation
The Defense Department is considering applying a “supply chain risk” designation to Anthropic. Such a designation is typically reserved for entities deemed to pose security or operational vulnerabilities within government supply chains.
If issued, the designation would have immediate and wide-ranging consequences:
-
Certification Requirement: All defense contractors working with the U.S. government would be required to certify that their systems and workflows contain no Anthropic technology.
-
Contractual Impact: Contractors unable or unwilling to remove Anthropic products could risk losing their Pentagon contracts.
-
Network Integration Challenges: Claude is currently the only commercial large language model (LLM) broadly approved and integrated within certain classified U.S. military networks. Its removal would require technical disentanglement from existing systems.
A senior Pentagon official acknowledged the logistical difficulty of removing the model from classified environments but indicated that the Department is prepared to proceed despite anticipated disruptions.
Financial and Strategic Implications
Anthropic holds a Pentagon prototype contract valued at approximately $200 million. While the direct financial impact of losing this contract may be limited relative to the company’s broader funding base, the broader consequences of a supply chain risk designation could extend beyond defense revenue.
Major commercial partners with government exposure may reassess their use of Anthropic’s technology to avoid jeopardizing their own federal contracts. This could affect integrations in enterprise software, data analytics platforms, and defense-adjacent technology services.
Claude is currently the only commercial LLM widely authorized for integration into certain classified U.S. defense networks. Transitioning to alternative AI systems would require additional certification, integration, and compliance processes.
Broader Context
The dispute underscores an emerging policy divide between segments of the U.S. technology sector and the Department of Defense over acceptable use standards for artificial intelligence in military contexts.
While several AI providers have shown flexibility in adapting consumer-level safety restrictions for defense contracts, Anthropic has maintained a framework that limits certain military and surveillance applications based on its internal governance policies and publicly stated safety commitments. The outcome of the Pentagon’s deliberations could shape future federal AI procurement standards and define how ethical guardrails are treated in government technology contracts.
As of this report, no final designation has been formally announced. The Department of Defense and Anthropic have not released additional official statements regarding the potential action.
——— End of Article ———
