Pentagon Drops Anthropic AI Tools After Supply Chain Risk Label
The Defense Department stopped using Anthropic's AI tools after blacklisting the company for refusing to strip safety guardrails from its products.
The Pentagon has ceased using artificial intelligence tools developed by Anthropic after designating the company a "supply chain risk," the BBC has learned. The decision marks a significant rupture between one of Silicon Valley's leading AI safety-focused firms and the U.S. military establishment.
The Defense Department applied the supply chain risk label to Anthropic in February, following the company's refusal to remove safety guardrails built into its AI systems. Those guardrails are designed to prevent the tools from producing harmful or dangerous outputs, a feature central to Anthropic's commercial and ethical identity.
Read more CFTC Investigates Adam Kinzinger Over Kalshi Pardon Bets →
The standoff highlights a deepening tension between government demands for unrestricted AI capability and the safety-first philosophy championed by a growing segment of the AI industry. Anthropic, founded by former OpenAI researchers, has positioned itself as a developer committed to responsible AI deployment — a stance that apparently conflicted with Pentagon requirements.
The blacklisting effectively bars Anthropic from serving as a vendor in that procurement relationship, a consequential outcome for a company that has actively sought government and enterprise contracts. The episode raises broader questions about how AI developers will navigate pressure from powerful institutional clients seeking fewer restrictions on the technology's behavior.
Continue reading at BBC News.
