The US Military Used Claude in the Iran Strikes Even After Trump Banned It
The United States military used Claude, Anthropic’s artificial intelligence model, during the large-scale strikes on Iran carried out alongside Israel, even after Donald Trump had ordered, just hours earlier, a complete break with the company. The information was reported by The Wall Street Journal and Axios, and it shines a light on a dilemma that goes far beyond politics or technology: what happens when an AI system is already so deeply woven into military operations that not even a presidential order can remove it overnight?
This episode is not an isolated case. It is the latest development in a crisis that began weeks earlier and involves ideological disputes, billion-dollar contracts, and a question that still has no easy answer: how far can governments rely on private companies to conduct war operations?
How Claude Was Used in Military Operations Against Iran
The relationship between Claude and the Pentagon did not start overnight. Anthropic’s artificial intelligence model had already been integrated into several US defense systems for months, with contracts in place and teams trained to operate the tool in highly complex scenarios. According to the reports, the US military command used Claude for intelligence purposes, in addition to helping with target selection and battlefield simulations.
In practice, Claude functioned as an advanced data analysis assistant. Military personnel and Pentagon analysts fed the system with information on troop movements, intercepted communications, and satellite imagery, and the artificial intelligence helped cross-reference it all at a speed impossible for human teams working alone. The AI processed information from multiple sources in real time, helping analysts identify patterns and make faster decisions during the joint operation with Israel.
It is important to point out that this was not an AI making autonomous decisions about targets or strikes. Claude dramatically accelerated the analysis and planning cycle, but final decisions remained in human hands. Even so, the level of integration was deep enough that disconnecting the tool from military systems in the middle of an active operation posed a real and potentially dangerous operational risk.
When the bombing of Iran began on Saturday, Pentagon teams were already using Claude as a central part of their workflow. The artificial intelligence was embedded in processes ranging from mission preparation to real-time monitoring. Swapping tools or simply shutting down the system at that moment would have been like asking a surgeon to change instruments in the middle of a critical procedure — technically possible, but extremely risky.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Trump’s Ban and the Origin of the Conflict With Anthropic
On Friday, just hours before the attacks began, Trump ordered all federal agencies to stop using Claude immediately. In a post on Truth Social, the American president described Anthropic as a radical left AI company run by people who have no idea what the real world is about.
But what triggered such an abrupt break? The conflict began with the use of Claude during the military operation to capture Venezuela’s president, Nicolás Maduro, in January. Anthropic publicly objected, pointing out that its terms of use prohibit Claude from being used for violent purposes, weapons development, or surveillance. That stance went over very badly at the White House and the Pentagon.
Since then, relations between Trump, the Pentagon, and Anthropic deteriorated rapidly. Defense Secretary Pete Hegseth posted a lengthy statement on X on Friday, accusing Anthropic of arrogance and betrayal. Hegseth was emphatic in saying that America’s warfighters would never be held hostage by the ideological whims of big tech companies. The secretary demanded full and unrestricted access to all of Anthropic’s AI models for any lawful purpose.
However, Hegseth also acknowledged how difficult it would be to quickly disconnect military systems from the tool, given how widely it was already being used. He said Anthropic would continue providing its services for a maximum of six months, allowing for a proper transition to what he called a better and more patriotic service.
The Clash Between Political Orders and Operational Reality
In practice, the Pentagon found itself in an extremely delicate situation. On one side, there was a clear presidential order to stop using Anthropic’s artificial intelligence. On the other, there was an ongoing military operation against Iran that depended in significant part on systems where Claude was already integrated.
Sources cited by the American outlets said senior officials chose to keep using the tool during the operation, understanding that the risk of shutting it down mid-process was greater than the risk of temporarily disobeying Trump’s directive. That decision alone raises a huge debate about the chain of command and what happens when technology becomes so essential that not even a presidential order can remove it instantly.
The episode also exposes a vulnerability that few people had been discussing until now: the growing dependence of the US military apparatus on artificial intelligence tools supplied by private companies. When the government hires a company like Anthropic and deeply integrates Claude into its operations, creating a viable alternative is not something that can be done in a matter of hours or days. There are contracts, training, system adaptations, and a learning curve that make the transition far more complex than simply signing an order.
The situation with Iran showed in practice, and in dramatic fashion, that artificial intelligence already occupies a strategic role in the armed forces that goes beyond what many people had imagined. We are no longer talking about pilot projects or lab experiments — we are talking about a technology that is already at the center of war operations.
OpenAI Steps In to Fill the Gap
With the break between Trump and Anthropic, rival OpenAI quickly positioned itself to take over the space left by Claude. Sam Altman, OpenAI’s CEO, announced that he had reached an agreement with the Pentagon to use the company’s tools, including ChatGPT, on the Department of Defense’s classified network.
This move by OpenAI is significant for several reasons. First, because the company has historically maintained restrictive policies on military use of its tools, but has been loosening that position in recent months. Second, because it shows how the artificial intelligence market focused on defense has become a competitive arena where billion-dollar contracts are at stake. Anthropic’s exit — if it actually happens after the six-month transition period — would represent a massive redistribution of resources and influence in the sector.
Beyond OpenAI, other companies such as Palantir, which already has an established relationship with the Pentagon, and startups focused on military AI applications are also watching this opportunity closely. The geopolitical landscape makes everything even more urgent. Tensions with Iran are unlikely to ease anytime soon, and the need for advanced artificial intelligence tools in defense operations is only expected to grow.
That means any company looking to replace Claude must not only have competitive technology, but also prove it can integrate quickly with existing military infrastructure. It is not enough to have the best language model on the market — it has to show it works under real operational conditions and under pressure.
The Ethical Dilemma of Artificial Intelligence in War Operations
Anthropic’s position in this story deserves attention. The company built its reputation around principles of safety and responsible AI use. Claude’s terms of use explicitly prohibit applications for violent purposes, weapons development, and surveillance. When the company discovered its tool was being used in the operation in Venezuela, it spoke out publicly against it — and that triggered the entire crisis that led to the break with the Trump administration.
This stance puts Anthropic in a complicated position. On one hand, the company is being consistent with its own principles and with its commitment to developing AI responsibly. On the other, it is walking away from massive US government contracts and leaving that market to competitors that may not share the same ethical concerns.
The question that remains is: who gets to define the limits of artificial intelligence use in military contexts? The company that develops the technology? The government that hires it? Or should there be broader regulation that clearly sets those boundaries? The case of Claude in the strikes on Iran shows that this regulatory gap is already producing real and potentially dangerous consequences.
What This Means for the Future of AI in Military Operations
The case of Claude at the Pentagon during the strikes on Iran will be studied for a long time as an example of what happens when political decisions and technological reality collide. Artificial intelligence is no longer just a supporting tool — it has become a structural component of modern military operations.
A few points became clear after this episode:
- Turning off an AI in the middle of a war is not as simple as pressing a button. The integration of these tools into military systems creates dependencies that take months to safely undo.
- Dependence on private companies for defense operations creates strategic risks. When a company can disagree with how the government uses its technology and threaten to cut it off, that creates a vulnerability that did not exist before.
- The military AI market is becoming increasingly contested and politicized. How companies position themselves on the use of their tools for warfare may determine who wins or loses billion-dollar contracts.
- The lack of clear regulation on AI in military contexts is a ticking time bomb. Without well-defined rules, high-impact decisions end up being made in the heat of the moment, based on political or commercial interests.
Regardless of who ends up taking Anthropic’s place at the Pentagon, one thing became obvious: the relationship between artificial intelligence and military power has entered a new phase, where the boundaries between technology, politics, and ethics are increasingly blurred. The episode involving Trump, Claude, and Iran is only the beginning of a debate that will define how these forces are balanced in the decades ahead — and it will require far more sophisticated answers than simply canceling a contract on Truth Social. 🤖
