What triggered the fallout?
Last July, Anthropic was awarded a $200 million Department of Defense contract to provide advanced AI for national security operations. Claude became the first frontier AI model cleared for use on a classified U.S. government network.
Tensions escalated when the Pentagon required Anthropic to sign a new agreement that would allow it to use Claude for “all lawful purposes.”
What does the Department of Defense actually want?
The Department of Defense argues that once a supplier signs a defense contract, it cannot dictate how its products are used.
Defense Secretary Pete Hegseth reportedly likened the Claude to military equipment, saying contractors have no control over how the jets or weapons systems are deployed.
Why don’t the Antropics agree?
Antropics does not completely deny military cooperation. It has worked closely with U.S. national security agencies and supported defense-related projects such as missile defense systems.
CEO Dario Amodei has repeatedly warned that the United States must stay ahead of China in the global AI race.
However, the company maintains two non-negotiable red lines:
-
There will be no large-scale domestic surveillance of American citizens.
-
There are no fully autonomous weapons that can select and engage targets without human supervision.
Anthropic allows Claude to support targeted attacks, foreign intelligence operations, and drone missions, as long as humans make the final decision.
Amodei argues that current AI models are not reliable enough to safely activate lethal autonomous weapons, and that mass surveillance using AI poses significant risks to civil liberties.
Flashpoint: Operation Venezuela
The conflict reportedly escalated after a January operation that led to the capture of Venezuelan President Nicolas Maduro. Claude was deployed via a platform operated by defense technology company Palantir Technologies.
Anthropic’s internal questions about how the model was used allegedly alarmed defense officials and deepened mistrust.
Mr. Hegseth later suggested publicly that the Pentagon would not work with AI systems that limit wartime deployment.
President Trump’s intervention
President Trump announced on Truth Social that he would direct all federal agencies to “immediately cease” the use of Anthropic’s technology.
He warned that if the companies were not cooperative during the phase-out period, he would use the full power of the “president” and could face civil and criminal consequences.
The General Services Administration suspended Anthropic from the USAi chatbot platform and began removing the company from federal procurement systems.
Despite the immediate order, the Pentagon reportedly has up to six months to phase out Claude, highlighting how embedded the technology is.
Legal threats and high-stakes pressure
The Pentagon set a deadline for compliance of 5:01 p.m. and threatened two major actions.
-
Invocation of the “Defense Production Act,” a Cold War-era law that allows the government to force private companies to prioritize defense contracts.
-
Designation of humanity as a “supply chain risk” — a label typically given to companies in hostile countries
Hegseth said contractors doing business with the U.S. military would be prohibited from doing business with Anthropic under such a designation.
Anthropic has vowed to challenge the supply chain risk label in court, calling the move an intimidation and a dangerous precedent.
Industry backlash and unity
The conflict sparked rare national unity within Silicon Valley.
More than 500 Google DeepMind and OpenAI employees have signed an open letter urging the companies not to capitulate to calls for mass surveillance and autonomous killing in the country.
Sam Altman, head of OpenAI, told Stuff that the company also opposes mass surveillance and autonomous weapons and is considering a contract with the Department of Defense to maintain similar safeguards.
The episode revealed a widening rift within technology. Some defense-minded investors and founders argue that AI companies should be allowed unrestricted military use, while others warn of the erosion of democratic safeguards.
paradox at the center
The U.S. government’s positions have attracted scrutiny because they appear contradictory.
As one former defense official put it, this technology is treated as existentially dangerous whether it is used or not.
What happens next?
If the Pentagon succeeds in enforcing compliance through legal pressure, commercial blacklisting, or emergency powers, a precedent could be established that U.S. AI companies cannot maintain their own safety restrictions contrary to government requirements.
If Antropic wins in court, it could strengthen the ability of private companies to put ethical guardrails in their defense agreements.
The dispute goes beyond one company and signals a turning point in relations between Silicon Valley and Washington. Although generative AI is primarily funded and controlled by private industry, it has become central to national security.
The outcome could not only determine the future of Anthropic, but also who ultimately decides how advanced AI is used for warfare and surveillance.
