Artificial intelligence company Anthropic has confirmed it will continue enforcing strict limits on military applications of its technology as negotiations with the U.S. defense establishment remain unresolved, according to a source familiar with the discussions.

The dispute centers on Anthropic’s refusal to relax safety safeguards that prevent its large language models from being used for autonomous weapons targeting or domestic surveillance operations. Defense officials have argued that compliance requirements should align with U.S. law and national security priorities.

During recent high-level talks, U.S. defense leadership warned that the government could take stronger action if the company does not adjust its policy. Options reportedly discussed included classifying Anthropic as a potential supply-chain security concern or invoking the Defense Production Act, a law that can compel corporate compliance in certain strategic sectors.

The Pentagon has been expanding partnerships with artificial intelligence developers to integrate advanced language models into defense operations, including cybersecurity monitoring, intelligence support, and battlefield data processing. The government is also exploring contracts with technology firms such as Google LLC, OpenAI, and xAI as part of broader modernization efforts.

Anthropic representatives stated that the company continues engaging in good-faith negotiations and aims to support national security objectives while preserving its responsible AI deployment framework. Company officials emphasized that current safety restrictions are designed to prevent misuse of AI systems in high-risk military scenarios.

The controversy gained attention after concerns emerged regarding potential use of AI tools during foreign military operations, although Anthropic said it had not raised specific operational complaints about third-party deployments.

Government officials also warned that if Anthropic is formally classified as a supply-chain risk, the decision could significantly affect the company’s commercial partnerships across the defense technology ecosystem. Legal experts from McCarter & English indicated that any adverse administrative action could trigger extensive litigation.

Defense officials have been working to shift AI contractors onto secure classified computing networks as part of a long-term modernization initiative. The Pentagon has already announced agreements with several AI developers to deploy models within controlled defense infrastructure environments.

The company’s safety stance reflects ongoing industry debates about the role of artificial intelligence in military automation. Advocates of strict safeguards argue that unrestricted deployment of AI systems in combat environments could introduce ethical and operational risks, particularly in autonomous decision-making scenarios.

Pentagon representatives have not publicly commented on the reported deadline given to Anthropic to respond to government demands by Friday evening.

Legal analysts believe the confrontation represents a broader policy challenge as governments worldwide attempt to balance rapid technological innovation with national security oversight. The outcome could shape future regulatory standards governing commercial AI systems used in defense research, intelligence processing, and operational planning.

For now, negotiations are expected to continue, with both sides attempting to avoid escalation while maintaining their respective strategic positions in the evolving artificial intelligence industry.