U.S. Bans Anthropic, Taps OpenAI for Pentagon
The Trump administration has banned federal agencies from using Anthropic's technology, designating the OpenAI rival a “supply chain risk” in an unprecedented move. The action came just hours before OpenAI finalized a major contract to provide its AI models to the Pentagon, effectively cementing its role as the government's preferred AI supplier.
The dispute between Anthropic and the Pentagon centered on the AI company's insistence on "red lines" that would prohibit the use of its technology for mass domestic surveillance or for autonomous weapons systems that can operate without human control. The Pentagon, however, insisted on the ability to use the technology for "all lawful purposes." The "supply chain risk" designation is a powerful tool typically reserved for foreign adversaries, such as China's Huawei. This marks the first instance of it being publicly applied to an American company. The designation could prevent tens of thousands of government contractors from using Anthropic's AI in their work for the Pentagon. OpenAI's new agreement with the Pentagon includes the very same prohibitions on mass surveillance and autonomous weapons that were the sticking point for Anthropic. OpenAI CEO Sam Altman stated the company also urged the Pentagon to offer these terms to all AI companies to "de-escalate" the situation. OpenAI's contract specifies that its models will be deployed via the cloud, with cleared OpenAI staff involved, and that the company retains control over its safety features. This conflict highlights a growing rivalry between OpenAI and Anthropic, which was founded in 2021 by former OpenAI employees, including former VP of research Dario Amodei. The split was reportedly due to disagreements over AI safety and OpenAI's direction. This rivalry has played out publicly, with Anthropic running a Super Bowl ad that was seen as a jab at OpenAI's plans to introduce ads to ChatGPT. The Trump administration's move has drawn criticism from some members of Congress. Democratic Senator Mark Warner and Representative Zoe Lofgren both expressed concerns that the decision may have been driven by political motives rather than a thorough analysis of national security. Anthropic has stated its intention to challenge the "supply chain risk" designation in court. The Pentagon has given federal agencies a six-month transition period to phase out their use of Anthropic's technology. The foundation of Anthropic's safety-focused approach is a technique developed by its researchers called "Constitutional AI." This method trains AI models using a set of principles, or a "constitution," to guide their behavior and align them with human values, reducing the need for constant human supervision to filter out harmful outputs. The controversy has sparked a debate within Silicon Valley, with nearly 500 employees from OpenAI and Google signing an open letter in support of Anthropic's position. The letter stated, "we will not be divided," and accused the Pentagon of trying to exploit competition between the companies.
