The American company Anthropic wanted to impose limits on the use of its Claude chatbot by the US Army, a measure that the Trump Administration refused. In retaliation, the US president ordered all federal agencies to stop using the technology of that firm, which will not be able to do business with the government or its suppliers. France 24 explains why these restrictions are essential and insufficient.

First modification: 02/28/2026 - 00:04

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 Pentagon wanted to deploy artificial intelligence (AI) across the Army, while Anthropic was less enthusiastic. The showdown between the US War Department and OpenAI competitor could determine the extent of the US military's use of chatbots and AI.

For Claude's creator, time has run out. Pete Hegseth, the Secretary of War, had ordered Anthropic to comply by this Friday, February 27 at 5:01 p.m. Eastern Time. Faced with the company's determination to stand firm, Donald Trump has urged his administration to stop using Claude.

"I am ordering ALL federal agencies of the United States government to IMMEDIATELY CEASE the use of Anthropic's technology. We do not need it, we do not want it and we will no longer work with them," declared the US president on his Truth Social platform.

To display this content from X (Twitter), you must allow audience measurement and advertising cookies.

The Republican clarified that agencies such as the War Department will have a six-month transition period to abandon the AI ​​company's services.

This sanction is not the only one. Hegseth has ordered the War Department to label Anthropic a "supply chain security risk," which "effective immediately" disqualifies it from any contract with the US government and prevents it from "any commercial activity" with "contractors, suppliers or partners doing business with the US Military."

Two red lines

In theory, the Pentagon is simply asking all AI vendors to allow the military to use chatbots "for all lawful purposes." However, Anthropic believes that this wording does not prevent the Army from crossing its two red lines in military matters.

The company does not want Claude to be used for large-scale surveillance of American citizens or for fully autonomous weapons, such as drones without human pilots.

The Anthropic rebellion may seem surprising. After all, the Army has been using Claude since last July. The agreement allows the integration of this chatbot into the US government's confidential communication networks and its use for "intelligence collection and processing," explains Sofia Romansky, coordinator of the research project on the responsible use of AI in the military field at the Center for Strategic Studies in The Hague.

However, Anthropic appears to have been surprised by the use of Claude in the context of the capture of Venezuelan President Nicolás Maduro in January. The company wanted to know more about how its AI was used during this controversial US military operation and whether its use fell within its ethical guidelines.

Shortly after this initial stumble in the relationship between the Pentagon and Anthropic, Pete Hegseth demanded the right to use these AIs "for all lawful purposes" in the Military.

The power struggle seeks to determine whether the State or private companies "control the legal and ethical frameworks for the military use of advanced data processing systems in generative AI models," summarizes Emil Archambault, researcher and specialist in emerging technologies in the military field at the German Council on Foreign Relations (DGAP).

An AI that presses the nuclear button?

The outcome of this confrontation becomes more important given the "significant change in the United States' position" regarding the military use of AI with the return of Donald Trump to the White House, says Anna Nadibaidze, a specialist in military applications of artificial intelligence at the University of Southern Denmark. The new administration "has redefined the concept of responsible use of AI, essentially stating that it would be irresponsible not to integrate this technology everywhere and without limits," explains this expert.

In other words, nothing would theoretically prevent Pete Hegseth and Donald Trump from allowing Claude or ChatGPT to decide when and against whom to use nuclear weapons.

Hence the importance of these 'red lines' that companies like Anthropic would like to be respected. They could fill a gap because "the legal frameworks are not yet defined and there are no clearly established standards that regulate how these technologies can be used by the military," Archambault notes.

The two limits established by Anthropic "represent the bare minimum," according to Ingvild Bode, director of the Center for War Studies at the University of Southern Denmark and an expert on the role of artificial intelligence in international security issues.

Although it is not much, "it is positive that a private company reminds us of these red lines," he adds.

They cover two crucial elements, according to experts interviewed by France 24. On the one hand, the refusal to allow Claude to participate in a program of mass surveillance of American citizens "affects a central ethical principle, respect for privacy," Nadibaidze emphasizes.

On the other hand, Anthropic does not believe that "current AI systems are capable of reliably performing the tasks necessary to control these autonomous weapons," explains Bode.

It is, therefore, a technical red line. These chatbots are still big black boxes, even for their creators. "They are made up of so many different parameters that there is great uncertainty about the final decision. Therefore, there are reliability problems that are difficult to reconcile with autonomous weapons," says Romansky.

For example, in the case of autonomous combat drones, no one can guarantee that the AI ​​"will be able to differentiate between a combatant and a civilian, or even distinguish an enemy soldier from an ally," he emphasizes.

"Insufficient" limits

In this context, these red lines are both a matter of principle and "a way to prevent possible legal risks," Bode emphasizes. Anthropic would not want to be held responsible for dramatic consequences on the ground. "If it were proven that an AI error caused the death of civilians, it could have very costly consequences for the company," adds Emil Archambault.

Although all the experts interviewed recognize the importance of these two red lines, "they are still insufficient," says Romansky. For example, the refusal to participate in a mass surveillance program "only applies to US citizens. What about foreign residents?" asks this specialist. This limit is also very vague: how many citizens participate before it can be considered mass surveillance?

The same goes for autonomous weapons. "The border between a 'fully' autonomous weapon and a 'partially' autonomous one is poorly defined," Archambault emphasizes. Even with human intervention, "we know that there is a tendency to rely too much on AI advice. When everything happens very quickly and humans are under pressure [as in the case of a drone attack, for example], they can be tempted to follow the algorithm's recommendations without necessarily verifying them themselves," explains Anna Nadibaidze.

Therefore, the problem is that, by establishing two red lines, as necessary as they are imperfect, Anthropic suggests that all other cases are acceptable. This is despite the fact that the example of semi-autonomous weapons suggests that the use of AI is problematic.

According to the experts interviewed, Anthropic's inflexibility regarding these principles has the advantage of exposing other AI giants who do not seem to share the same reservations.

The Pentagon can still rely on ChatGPT (OpenAI), Grok (Elon Musk's xAI), and Gemini (Google), if Anthropic doesn't budge. However, this impasse seems to have raised some consciences: more than 100 Google employees have signed an open letter urging their leaders to follow Anthropic's example.

This battle could also leave its mark if, faced with Pete Hegseth and Donald Trump, Anthropic decides to give in after all. In this case, "smaller companies, which also wanted to take an ethical stance, might conclude that it doesn't make sense because they will never be able to resist government pressure," says Nadibaidze.

The entire American AI ecosystem would then risk bowing to the Trump administration's demands.

This article was adapted from its original in French

Related keywords

The content you requested does not exist or is no longer available.