WASHINGTON — A crucial meeting is scheduled for Tuesday between Defense Secretary Pete Hegseth and the head of artificial intelligence company Anthropic, as disagreements continue over how AI should be deployed in military operations.
Anthropic, the company responsible for creating the Claude chatbot, stands alone among major AI firms in refusing to participate in the Pentagon’s new internal technology network. CEO Dario Amodei has expressed serious ethical reservations about unrestricted government AI applications, particularly regarding fully automated weaponized drones and AI-powered mass monitoring systems that could track political opposition.
A defense department source, speaking anonymously due to lack of authorization for public statements, verified the upcoming discussion between Hegseth and Amodei.
This scheduled conference highlights ongoing tensions surrounding artificial intelligence’s place in national defense and worries about the technology’s potential use in life-or-death scenarios, classified intelligence matters, or government monitoring programs. The meeting also occurs as Hegseth has promised to eliminate what he describes as “woke culture” within military ranks.
“A powerful AI looking across billions of conversations from millions of people could gauge public sentiment, detect pockets of disloyalty forming, and stamp them out before they grow,” Amodei wrote in an essay last month.
Last summer, the Pentagon revealed it was granting defense agreements to four artificial intelligence companies — Anthropic, Google, OpenAI, and Elon Musk’s xAI. Each agreement carries a potential value of $200 million.
Anthropic became the initial AI firm to receive approval for classified military systems, where it collaborates with partners including Palantir. The remaining three companies currently function only within unclassified settings.
By the start of this year, Hegseth was promoting just two companies: xAI and Google.
During a January address at Musk’s SpaceX facility in South Texas, the defense secretary stated he was dismissing any AI systems “that won’t allow you to fight wars.”
Hegseth outlined his military AI vision as systems functioning “without ideological constraints that limit lawful military applications,” then emphasized that Pentagon “AI will not be woke.”
In January, Hegseth announced that Musk’s AI chatbot Grok would become part of the Pentagon system called GenAI.mil. This declaration followed shortly after Grok — integrated into Musk’s social platform X — faced worldwide criticism for creating explicit deepfake images without individuals’ permission.
OpenAI revealed in early February that it would also participate in the military’s protected AI platform, allowing service personnel to access a specialized ChatGPT version for non-classified duties.
Since its establishment in 2021 by former OpenAI executives, Anthropic has consistently marketed itself as the most responsible and safety-focused among leading AI companies.
The Pentagon situation is testing those commitments, according to Owen Daniels, associate analysis director and fellow at Georgetown University’s Center for Security and Emerging Technology.
“Anthropic’s peers, including Meta, Google and xAI, have been willing to comply with the department’s policy on using models for all lawful applications,” Owens said. “So the company’s bargaining power here is limited, and it risks losing influence in the department’s push to adopt AI.”
Following ChatGPT’s launch and the resulting AI enthusiasm, Anthropic worked closely with President Joe Biden’s team by volunteering to allow independent evaluation of its AI systems to prevent national security threats.
CEO Amodei has cautioned about AI’s potentially devastating risks while rejecting characterization as an AI “doomer.” In his January essay, he contended that “we are considerably closer to real danger in 2026 than we were in 2023” but argued those threats should be addressed in a “realistic, pragmatic manner.”
This wouldn’t mark the first occasion Anthropic’s push for enhanced AI protections has created friction with the Trump administration. Anthropic publicly challenged chipmaker Nvidia, condemning Trump’s proposals to relax export restrictions that would permit certain AI computer chips to be sold to China. Despite this criticism, the AI company maintains a strong partnership with Nvidia.
The Trump administration and Anthropic have also found themselves on opposing sides regarding AI regulation efforts in various states.
David Sacks, Trump’s senior AI advisor, claimed in October that Anthropic was “running a sophisticated regulatory capture strategy based on fear-mongering.”
Sacks posted these comments on X responding to Anthropic co-founder Jack Clark’s discussion about balancing technological optimism with “appropriate fear” regarding the continuous development of more advanced AI systems.
Following Trump’s White House return, Anthropic recruited several former Biden administration officials while attempting to demonstrate bipartisan appeal. The company recently appointed Chris Liddell, a former White House official from Trump’s initial presidency, to its board.
The Pentagon-Anthropic disagreement echoes a previous controversy when technology workers protested their companies’ involvement in Project Maven, a Pentagon drone monitoring program. Although some employees resigned over the initiative and Google withdrew, the Pentagon’s dependence on drone surveillance has continued expanding.
Similarly, “the use of AI in military contexts is already a reality and it is not going away,” Owens said.
“Some contexts are lower stakes, including for back-office work, but battlefield deployments of AI entail different, higher-stakes risks,” he explained, referencing lethal force applications or weapons systems like nuclear arsenals. “Military users are aware of these risks and have been thinking about mitigation for almost a decade.”
