Why Trump administration is pressuring Anthropic over military AI?
What's the story
US Defense Secretary Pete Hegseth has threatened to cut Anthropic off from his agency's supply chain if the company does not allow its artificial intelligence (AI) technology to be used in military applications. The threat was made during a Pentagon meeting with Anthropic CEO Dario Amodei on Tuesday. A senior Pentagon official said that Anthropic has until Friday evening to comply with the demand.
Company position
Anthropic responds to ultimatum
In response to Hegseth's ultimatum, Anthropic said they had "continued good-faith conversations about our usage policy." The company emphasized its commitment to support the government's national security mission within the scope of what their models can reliably and responsibly do. However, a senior Pentagon official clarified that this dispute does not involve autonomous weapons or mass surveillance.
Legal implications
Potential invocation of Defense Production Act
If Anthropic fails to comply with Hegseth's demand, the Defense Production Act could be invoked against the company. This legal measure could force Anthropic executives to grant unrestricted access by the Pentagon on national security grounds. The Pentagon would also designate Anthropic as a supply chain risk during this process.
Company boundaries
Ethical boundaries set by Anthropic
Anthropic has made it clear that it will not allow its tools to be used for autonomous kinetic operations where AI tools make final military targeting decisions without human intervention. The company also objects to the use of its tools for mass domestic surveillance. Despite these boundaries, the Pentagon remains adamant that Anthropic should have no say in how its products are used by the military.
Model utilization
Dependency on Claude complicates situation
Anthropic's Claude is currently the only AI model integrated into the Pentagon's most sensitive classified systems. This strategic dependency complicates any threat to terminate their relationship. The model is said to be used in highly sensitive operational contexts.
Strategic shifts
Alternatives being explored by Pentagon
The Pentagon is said to be fast-tracking talks with OpenAI and Google to bring their models into classified environments. Gemini has emerged as a potential alternative, but such an arrangement would require Google to allow the Pentagon to use its system for "all lawful purposes," terms that Anthropic has declined. Elon Musk's xAI recently signed a contract to bring Grok into classified settings, but it remains unclear if this system could fully replace Claude's current functions.