The artificial intelligence company Anthropic has until February 27 to withdraw safeguards and regulations on the use of its tools by the Pentagon, according to 'Axios', which reports that the deadline was set by US Secretary of Defense Pete Hegseth to Anthropic's executive director Dario Amodei, in a new chapter of tensions between the company and the Government of Donald Trump.
First modification: 02/24/2026 - 22:38
United States Secretary of Defense Pete Hegseth set a deadline of February 27 for Dario Amodei, CEO of the company Anthropic, to give the Pentagon unrestricted access to its artificial intelligence model, which the contractor has previously refused.
According to sources who spoke with the 'Axios' portal, in a meeting with the businessman, Hegseth would have included in the ultimatum the threat of strong sanctions if the AI company refuses the request.
The information was also reproduced by the Reuters agency, which had anticipated a non-friendly meeting, although a company spokesperson told 'Axios': "We are having productive conversations, in good faith."
Defense officials who testified to 'Axios' indicated that such retaliation could range from cutting commercial ties to declaring the company a "supply chain risk," to the application of the Defense Production Act, which would force Anthropic to adjust its model to the Pentagon's demands.
Anthropic is one of four AI companies to which the Pentagon awarded defense contracts worth more than $200 million each last year, along with Google, OpenAI and Elon Musk's xAI.
But it is the only one that operates in a classified environment, essential for the management of sensitive information such as military information.
A source familiar with the meeting who spoke with the Reuters agency said that the company has no intention of relaxing its restrictions, despite pressure from the Pentagon.
Difference of visions
Anthropic's chatbot, Claude, is believed to have been used in both the planning and execution of the capture of President Nicolás Maduro during a daring US Delta Force raid on Caracas on January 3, a use that could have been a critical source of disagreements between the Pentagon and Anthropic.
According to a senior Trump Administration official cited by 'Axios', Anthropic requested information on whether Claude had been employed during the operation, a version that coincides with the corporate effort to learn the conditions of use of its technology. That would have unleashed the annoyance of the War Department.
“Any company that jeopardizes the operational success of our fighters on the ground is a company with which we must reevaluate our collaboration in the future,” the source told 'Axios'.
An Anthropic spokesperson denied that any consultations had taken place “outside of routine conversations about strictly technical matters.”
Anthropic has positioned itself as a leading AI company that prioritizes the security and use of its tools in accordance with specific regulations.
These restrictions have been maintained since the presidency of Joe Biden, who considered it necessary to guarantee the responsible use of a technology with potential for cyberattacks, mass surveillance or the development of autonomous lethal devices.
The door closed to 'woke' contractors
Shortly before his departure from the White House, Biden enacted a framework that prohibited applications of AI systems that could violate constitutionally protected rights or automate the deployment of nuclear weapons.
These regulations remain in force in the corporate policies of Anthropic, which refuses to open its tools to the Pentagon without applying the same red lines that they impose for other users: that its technology is not used to develop autonomous weapons or for espionage on American citizens.
"A powerful AI that analyzes billions of conversations from millions of people could measure public sentiment, detect pockets of disloyalty forming, and eradicate them before they grow," Anthropic CEO Dario Amodei wrote in an essay published in January.
The contractor's demands have infuriated Hegseth, who has defended that AI systems must operate "without ideological restrictions that limit legal military applications" and has insisted that under his management the “Pentagon's AI will not be woken up.”
To display this YouTube content, you must allow advertising and audience measurement cookies.
An extension on your browser appears to be blocking the video player from loading. In order to view this content, you must disable it on this site.
The Trump administration's top AI advisor, David Sacks, has accused Anthropic of "carrying out a sophisticated regulatory capture strategy based on fear-mongering."
Precisely as an alternative to progressive policies, Elon Musk developed the Grok bot, which operates embedded in X and is used by the Pentagon. At the time, it sought to oppose interactions with other AI chatbots such as Google's Gemini or OpenIA's ChatGPT, which in Musk's opinion were permeated by 'woke' thinking.
But Grok hasn't exactly been a reliable experience. In its beginnings it caused controversy with anti-Semitic comments and praise for Adolf Hitler and more recently it generated scandal for the creation of hypersexualized images of women and even girls.
With Reuters, AP and local media
Related keywords
The content you requested does not exist or is no longer available.




