In response to being listed as a "supply chain risk entity" by the U.S. Department of Defense (DOD), Anthropic CEO Dario Amodei made an official statement on Thursday. He announced the company's plan to challenge the decision in court, calling it legally "untenable."

Before this designation, there had been a weeks-long struggle between the two sides over military control of AI systems. According to the restrictions of this label, Anthropic might be prohibited from doing business with the Pentagon and its contractors. Amodei reiterated that the company remains committed to not using AI for mass surveillance or creating fully autonomous weapons, but the Department of Defense seeks unrestricted access for "all legitimate uses."

Amodei emphasized that this designation mainly affects customers with direct contracts with the Department of the Army, and has little impact on the majority of general customers. He pointed out that the law requires relevant departments to take "least restrictive" measures to ensure supply chain security, while the current designation is too broad and lacks sufficient basis.

Additionally, Amodei publicly apologized for previously leaked internal emails. In those emails, he had evaluated OpenAI's deals with the Department of Defense as "a security show." He explained that the email was written during a "difficult time" when the company faced multiple pressures, and represented an emotional, outdated assessment, not his considered position.

Currently, Anthropic's top priority is ensuring frontline personnel can still use essential tools. To this end, the company has promised to continue providing model support to the Department of Defense at "nominal cost" until the transition is complete.

Key Points

  • ⚖️ Legal Counterattack: Anthropic believes the Department of Defense's designation lacks legal basis and is prepared to overturn the label through judicial procedures.

  • 🛡️ Sticking to Principles: Clearly prohibiting AI from being used for autonomous weapons and surveillance creates a core conflict with the Department of Defense's demand for "unrestricted access."

  • 💰 Transition Period Guarantee: To avoid affecting national security operations, the company will continue to provide technical support at symbolic prices during the transition period.

  • 📝 Apology and Correction: The CEO withdrew negative comments about OpenAI, trying to ease the public relations storm caused by the leak of internal emails.