A 2011 file photo of the Pentagon, the headquarters of the Department of Defense.
STAFF/AFP via Getty Images
hide caption
toggle caption
STAFF/AFP via Getty Images
Defense Secretary Pete Hegseth is threatening to bar Anthropic from collaborating with the U.S. military due to the company’s steadfast commitment to its safety protocols.
This warning emerged on Tuesday during a meeting between Hegseth and Anthropic CEO Dario Amodei, as recounted by two individuals familiar with the discussions who requested anonymity.
For several months, Amodei has maintained that the implementation of AI for domestic surveillance and AI-operated weapons crosses ethical boundaries, labeling such practices as “illegitimate” and “prone to misuse.” A source who attended the Hegseth meeting confirmed that Amodei reiterated these views on Tuesday.
Hegseth has insisted that Anthropic must permit the application of its AI tools for all “lawful” purposes, which may encompass warfare and surveillance directed by AI.
Pentagon officials have indicated that the Defense Department intends to continue utilizing Anthropic’s resources, irrespective of the company’s stance.
If Anthropic remains unyielding, Hegseth is prepared to invoke the Defense Production Act, a law initially enacted in the 1950s to compel companies to produce essential goods during national emergencies.
This action would mandate Anthropic to permit military use of its tools, whether they agree or not, according to a senior Pentagon official.
The response from the Trump administration could also involve categorizing Anthropic as a “supply chain risk,” effectively placing the company on a government blacklist, per the sources acquainted with the meeting.
Anthropic CEO Dario Amodei at a meeting in New Delhi last month.
Ludovic Marin/AFP via Getty Images
hide caption
toggle caption
Ludovic Marin/AFP via Getty Images
Hegseth, along with other officials from the Trump administration, has labeled Anthropic’s firm stance against domestic surveillance and AI weapons as “woke AI.” White House AI advisor David Sacks was involved in drafting a recent executive order aimed at tech companies regarding this issue.
Experts in AI describe “woke AI” as a vague term used by Trump officials to label any safety measures associated with advanced AI tools, as well as the belief that AI chatbots exhibit liberal biases in their programming.
In contrast, other AI organizations like OpenAI and Google have agreed to permit the use of their technologies in any “lawful” contexts, as has Elon Musk’s xAI, which recently received approval for classified operations.
Last summer, the Pentagon granted contracts worth up to $200 million each to Anthropic, Google, OpenAI, and xAI, with Anthropic being the first approved for classified operations due to its status as the most advanced and secure option for sensitive military needs.
The ongoing tension with the White House arises at a pivotal moment for Anthropic, which is preparing for an initial public offering later this year. It remains uncertain how this discord with the administration will affect investor sentiments.
Amodei has highlighted that Anthropic’s valuation and revenue have continued to rise since the company took a principled stance against the Trump administration regarding the deployment of AI in military settings.
“My primary concern is that a small number of individuals could ultimately control a drone army without requiring human collaboration,” Amodei expressed in a January essay outlining his apprehensions. “I believe we should approach fully autonomous weaponry with extreme caution and not advance its use without adequate safeguards.”
