Anthropic, the artificial intelligence company behind the Claude chatbot, has seen its relationship with the Pentagon deteriorate over disagreements about safety guardrails on its technology.
According to reporting summarized by MSN, emails reveal that Undersecretary Emil Michael and Anthropic CEO Dario Amodei went back and forth for months over those safety guardrails before the relationship fell apart. The exchange, described as playing out over an extended period, centered on the limits Anthropic places on how its AI can be used.
According to Fortune, Anthropic has "bucked the rules of Trump's Washington" — and the company has paid a price for doing so. Fortune frames the episode as a case study in what happens when a firm resists the expectations of the current administration.
Taken together, the two accounts describe a standoff: a leading AI developer that has built its brand around caution and safety limits on one side, and defense officials seeking access to that technology on the other. The disagreement over guardrails appears to have been the sticking point that soured the partnership.
Beyond these points, the sources do not detail the specific contract, the exact guardrails in dispute, or the financial terms involved.
Why it matters: As the U.S. military increasingly seeks to adopt commercial AI, this breakdown highlights the growing friction between companies that want to restrict how their tools are used and a government that wants fewer limits — a tension that will shape who supplies the technology behind national defense.