The Pentagon has halted its use of artificial intelligence tools developed by Anthropic following the company's placement on a blacklist, according to BBC reporting. The U.S. Department of Defense designated Anthropic a "supply chain risk" in February after the company declined to remove safety guardrails from its AI systems.
Safety guardrails are built-in limitations designed to prevent AI systems from generating harmful, unethical, or dangerous content. Anthropic's refusal to strip these protections from its tools prompted the Pentagon's decision to restrict access.
The blacklisting represents a significant tension between national security interests and AI safety practices. The Pentagon has not publicly detailed which specific tools or systems it stopped using or the scope of the previous deployment within the department.
Anthropic, founded in 2021 by former OpenAI researchers, has positioned safety as a core principle of its business model. The company develops Claude, a large language model designed with multiple layers of safety features. The firm's approach contrasts with some pressure from defense and intelligence agencies to maximize AI capabilities without restrictions that could limit operational flexibility.
The BBC report indicates the Pentagon views the company's refusal to compromise on safety measures as a business risk rather than acknowledging potential benefits of maintaining robust safety standards. This decision affects Anthropic's access to one of the U.S. government's largest technology procurement operations.
It remains unclear whether other federal agencies have taken similar actions against Anthropic or whether this represents a Pentagon-specific determination. The company has not publicly commented on the blacklisting or confirmed the specific reasons the Pentagon cited.