The Pentagon and Anthropic are in a standoff over who controls the use of AI in military operations, with the government threatening to force Anthropic to provide its AI tools if the company doesn’t comply with defense demands. This conflict highlights the broader struggle between government authority and private tech firms over the deployment, safety, and ethical boundaries of artificial intelligence in warfare.
The Pentagon is intensifying its dispute with Anthropic, an AI company, over the control and deployment of artificial intelligence in modern warfare. The conflict centers on whether the U.S. government or private tech firms should dictate how AI is used in military operations. Anthropic’s CEO, Dario Amodei, has been given a strict deadline to respond to an ultimatum from the Department of Defense. Defense Secretary Pete Hegseth has threatened to invoke the Defense Production Act, a law dating back to the Korean War, which would allow the government to compel Anthropic to provide its AI tools under wartime authorities if the company refuses to cooperate.
The Pentagon’s concerns are rooted in hypothetical but urgent scenarios, such as a nuclear missile attack where only seconds are available to respond. Officials worry that Anthropic’s AI safeguards could prevent a timely, AI-assisted response in such critical moments. In response to these concerns, Anthropic has made some concessions, allowing its AI to be used in missile defense and cyber operations. However, the company maintains strict boundaries, refusing to permit its technology on classified networks, for autonomous drone targeting of enemy combatants, or for mass surveillance of U.S. citizens.
The Department of Defense insists that humans will always be involved in decisions to use force and that all actions will remain lawful. However, Pentagon leaders argue that they cannot allow a private company’s usage policies to dictate battlefield decisions. If Anthropic does not comply with the government’s demands, it risks losing up to $200 million in Pentagon contracts. Additionally, it could be labeled a supply chain risk, which would force major defense contractors like Palantir and Lockheed Martin to certify that they are not using Anthropic’s AI models.
This standoff comes at a time when Anthropic, recently valued at $38 billion, is adjusting its own safety framework. The company has announced it will no longer delay the release of powerful AI models if competitors such as OpenAI and xAI are moving faster, signaling a shift toward a more aggressive stance in the AI race. This change reflects the broader pressure within Silicon Valley to keep pace with rivals, even as concerns about safety and ethical use remain unresolved.
Ultimately, the clash between Anthropic and the Pentagon highlights a fundamental question about the future of AI in warfare: who should control these powerful technologies, and how can their use be made both effective and safe? The U.S. government is pushing for an “AI-first” military, while tech companies are eager to lead the global AI race. The outcome of this dispute will set important precedents for the balance of power between state authority and private innovation in the age of artificial intelligence.