US Bans Anthropic AI, Taps OpenAI
The Trump administration has abruptly banned AI-startup Anthropic from all federal and military use, citing "supply chain risk." Within hours, competitor OpenAI secured a major Pentagon contract to fill the void, providing its most advanced models with new "technical safeguards" to address government safety concerns.
The abrupt banishment of Anthropic stemmed from a months-long, contentious negotiation with the Pentagon over the military's use of its AI model, Claude. Anthropic insisted on contractual "red lines" to prevent its technology from being used for mass domestic surveillance or in fully autonomous weapons systems that can kill without human input. The Trump administration and Defense Secretary Pete Hegseth demanded "full, unrestricted access" to Anthropic's models, arguing that such corporate-imposed constraints were operationally unrealistic and impeded national security. When Anthropic CEO Dario Amodei refused to concede, President Trump labeled the company "Radical Left, Woke" and ordered the cessation of all federal use of its technology. The "supply chain risk" designation is a severe and unprecedented action against a U.S. company, a measure historically reserved for foreign adversaries. This designation could prevent any U.S. military vendor from working with Anthropic, potentially crippling the company's business ahead of an anticipated IPO and impacting partners like Amazon Web Services and Palantir. OpenAI, which had previously excluded military uses in its terms, has recently pursued government contracts more actively through its 'OpenAI for Government' initiative. The new Pentagon agreement contains language prohibiting the use of its AI for mass domestic surveillance or to independently direct autonomous weapons where human control is required by law or policy. OpenAI states its agreement has more guardrails than Anthropic's, citing a multi-layered approach that includes retaining control over its safety stack, cloud-only deployment, and having cleared OpenAI personnel involved. CEO Sam Altman suggested Anthropic may have sought more operational control than OpenAI was willing to concede. The Pentagon had awarded both Anthropic and OpenAI contracts with ceilings of $200 million to prototype AI capabilities. Anthropic's Claude was reportedly the only frontier AI tool available for U.S. government users on networks for information classified up to the secret level. The Pentagon now has a six-month period to phase out Anthropic's embedded technology.
