Anthropic Statement on Department of War AI Deployment and Safeguards

Anthropic maintains safeguards against mass domestic surveillance and fully autonomous weapons

Anthropic is refusing to remove specific safety guardrails regarding mass domestic surveillance and fully autonomous weapons, despite threats from the U.S. Department of War to designate the company as a "supply chain risk" or invoke the Defense Production Act. The company maintains that these safeguards are necessary to protect democratic values and ensure the safety of warfighters and civilians.

Deployment of Claude in National Security

Anthropic has proactively integrated its Claude models into U.S. government classified networks, National Laboratories, and national security agencies. The company was the first frontier AI company to achieve these deployments and provides custom models specifically for national security customers.

Claude is currently used for several mission-critical applications, including:

Intelligence analysis
Modeling and simulation
Operational planning
Cyber operations

National Security and Export Controls

To maintain a democratic advantage in AI, Anthropic has taken several measures that prioritize national security over short-term revenue:

Revenue Forfeiture: The company forgave several hundred million dollars in revenue by cutting off Claude's access to firms linked to the Chinese Communist Party (CCP), including entities designated as Chinese Military Companies by the Department of War.
Cyber Defense: Anthropic shut down CCP-sponsored cyberattacks targeting Claude.
Policy Advocacy: The company has advocated for strong export controls on chips to prevent autocratic adversaries from gaining an AI lead.

Specific Use Case Restrictions

Anthropic states that while it does not interfere with military decisions or specific operations, it maintains two strict exceptions in its contracts with the Department of War:

Mass Domestic Surveillance

Anthropic opposes the use of AI for mass domestic surveillance, arguing it is incompatible with democratic values. The company asserts that AI can automatically assemble scattered, individually innocuous data—such as movement records, web browsing, and associations—into a comprehensive picture of a person's life at a massive scale, presenting novel risks to fundamental liberties.

Fully Autonomous Weapons

Anthropic refuses to power fully autonomous weapons—systems that automate the selection and engagement of targets without human intervention. The company cites two primary reasons for this restriction:

Reliability: Frontier AI systems are currently not reliable enough to power such weapons without putting warfighters and civilians at risk.
Lack of Oversight: Without proper guardrails and oversight, these systems cannot exercise the critical judgment exhibited by professional troops.

Anthropic has offered to collaborate with the Department of War on R&D to improve system reliability, but the offer was not accepted.

Conflict with the Department of War

The Department of War has demanded that Anthropic accede to "any lawful use" and remove the aforementioned safeguards. Failure to comply has led to the following threats from the Department:

Removal of Anthropic from government systems.
Designation as a "supply chain risk," a label typically reserved for U.S. adversaries.
Invocation of the Defense Production Act to force the removal of safeguards.

Anthropic describes these threats as contradictory, noting that the company is simultaneously being labeled a security risk while its technology is viewed as essential to national security. Despite these threats, Anthropic states it cannot "in good conscience" remove the safeguards. If the Department chooses to offboard the company, Anthropic has committed to ensuring a smooth transition to another provider to avoid disrupting military planning or critical missions.