Anthropic CEO Dario Amodei said the company would not lift safeguards from its AI system Claude despite pressure from the Pentagon, warning that some uses of artificial intelligence could undermine rather than protect democratic values.
“In any event, these threats do not change our position. We cannot in good conscience comply with their demands,” Amodei said in a detailed statement outlining the company’s position.
The dispute centers on the Pentagon’s requirement that AI contractors agree to “all lawful uses” of the system, including potential applications that Anthropic crosses ethical and technological boundaries.
AI for defense — limited
Amodei revealed that Anthropic is a strong supporter of deploying AI to strengthen U.S. national security.
“I believe deeply in the existential importance of leveraging AI to protect the United States and other democracies and defeat authoritarian adversaries,” he said.
Anthropic is the first frontier AI company to deploy models on sensitive U.S. government networks and national laboratories. The company says Claude is used across the defense and intelligence community for intelligence analysis, operational planning, cyber operations, and modeling.
Amodei said Antropic also cut off access to companies with ties to the Chinese Communist Party, losing hundreds of millions of dollars in revenue in the process.
two red lines
The current standoff revolves around two specific use cases that Anthropic denies permission for: domestic mass surveillance and fully autonomous weapons.
Regarding surveillance, Amodei said large-scale surveillance of Americans using AI poses serious risks.
“The use of these systems for domestic mass surveillance is inconsistent with democratic values,” he wrote. He warned that AI could “automatically and at scale” combine vast amounts of publicly available data such as browsing records, movement data, and relationships into a comprehensive picture of an individual’s life.
Regarding fully autonomous weapons, Amodei acknowledged that they could someday play a role in national defense. But he said current Frontier AI systems “simply aren’t reliable enough” to remove humans from targeting decisions entirely.
“We do not knowingly provide products that endanger U.S. warfighters or civilians,” he said.
Anthropic offered to work with the ministry on research and development to improve reliability, but Amodei said the offer was not accepted.
pentagon pressure
Amodei said the Pentagon threatened to remove Anthropic from its systems if safeguards were maintained. Officials also reportedly warned that they could designate the company as a supply chain risk and invoke the Defense Production Act to force restrictions to be lifted.
Defense Secretary Pete Hegseth on Tuesday gave Anthropic an ultimatum: open up its artificial intelligence technology to unrestricted military use by Friday or risk losing government contracts.
Amodei said these threats are contradictory.
“Some label us a national security risk, while others label Claude essential to national security,” he wrote.
Amodei said that despite the tensions, Anthropic wants to continue its service to the department. If offboarded, the company will work to ensure a smooth transition to avoid disruption to military operations.
“We stand ready to continue our efforts to support the national security of the United States,” he said.
– end
Please tune in
