NoHe’s not a general, he’s not a weapon, he’s not a missile system. Yet his name has become the heart of one of the most delicate discussions of the present, the relationship between artificial intelligence, war and political power. That name It’s Claude, an AI system developed by American company Anthropicended up in the center of a political and military clash with the United States government and Donald Trumpwhich concerns national security and the use of artificial intelligence in military and classified systems. And not least the attacks in Iran.
Claude, the “different” artificial intelligence
To understand why software has become a matter of geopolitical conflict, we must first understand what Claude really is. It is a generative artificial intelligence model, that is, a capable system to understand complex texts, synthesize large amounts of informationproduce analyses, simulate scenarios, process volumes of data in just a few seconds that would require weeks of human work.
The overly ethical AI that annoys Trump
To develop it is Anthropic, a company founded by former OpenAI researchers (creators of ChatGPT) who, since its birth, wanted its AI to stand out for its policy of “safety first”, imposing strict limits on how its technology can be used, especially by the government and the armed forces. And it is precisely these limitations that made Claude a delicate political case, particularly in negotiations with the Department of Defense, which asked for freer use of the system.
The Department of Defense wants total freedom in the use of AI, Anthropic sets limits. Trump enters the scene with a post. (Getty Images)
When artificial intelligence enters military systems
In recent years, in fact, adapted versions of Claude they have been integrated into US government environments, including on classified networks. Not to pull a trigger, but to analyze information: intelligence reports, interceptionssatellite images, strategic scenarios. In a context of growing tension, these tools become invaluable for operational planning and to accelerate human decision-making. But while the Pentagon wants to be able to use AI without constraints, Anthropic believes some uses of AI are too risky.
The rules don’t change
The conflict exploded when the Department of Defense has called for more freedom in using the technologyincluding access to commercial data on American citizens for mass analysis. Essentially, he asked Anthropic to eliminate some restrictions to allow wider uses. Which the company rejected, saying that artificial intelligence is not mature enough to operate in contexts of mass surveillance or autonomous weapons without risks for civilians and soldiers.
Claude and the clash with the Pentagon
The political response was immediate. The Department of Defense has designated Anthropic as “risk in the supply chain”, a definition usually reserved for foreign entities perceived as a threat to national security. Trump entered the scene with a straight leg with a post on his social network Truth in which he orders the company to be canceled from federal contracts, transforming a technical negotiation into a national political case. Contextually and inevitably the negotiations were interrupted: although an agreement somehow seemed close, philosophical differences, personal frictions and rivalries caused the agreement to founder.
Claude used for attacks on Iran
According to what was reported by British newspaper The Guardian, the US Army would still use Claude to support military operationsi during the massive joint bombing conducted by Washington and Tel Aviv against Iran, starting on Saturday 1 March 2026.
A column of smoke rises from a neighborhood in Tehran after the area was targeted in attacks on March 1, 2026. The Islamic Republic of Iran Broadcasting (IRIB) building after Iranian authorities said it was targeted in attacks, while the Iranian military announced it had launched new attacks against US and Israeli targets (Photo by Fatemeh Bahrami/Anadolu via Getty Images)
According to the newspaper’s reconstruction, the AI system would be employed in different phases of the offensive: from the analysis of data and strategic intelligence information to the selection of potential targets, up to the conduct of tactical simulations of battlefield operations. A use that signals a now advanced level of integration of artificial intelligence technologies in the Pentagon’s operational activities.
The most controversial element, he highlights The Guardianis that Claude’s employment would take place despite President Donald Trump ordering all federal agencies to suspend their use just hours before the attack begangoing so far as to call Anthropic a “radical left AI company.”
The fact that the tool was still used suggests, according to the newspaper, how the technology was already deeply rooted in US military missions and decision-making flows, making immediate disengagement impracticable. A circumstance that contrasts with the positions expressed in the past by Anthropic, which in its terms of service had expressed strong objections to the use of its models for violent purposes, surveillance or in the development of armaments.
The paradox of technological dependence
There is in fact something that perhaps Trump ignores: despite the political threats, in fact, the Claude system cannot be removed overnight from federal systems, being already very integrated into complex workflows, operational analyzes and decision-making systems. And this creates an inevitable paradox: while his ban is being discussed, the technology continues to be used. The story demonstrates how artificial intelligence is no longer a simple experimental tool, but a structural component of the modern decision-making apparatus.
The ethical question: who decides the limits?
At the center of the controversy is not just software, but a question destined to shape the next decade: Who sets the limits of artificial intelligence when it enters the military sphere? The State claims it must have full control of defense instruments. Technology companies, however, own the know-how, the code and the infrastructure. And some, like Anthropic, claim the right to impose restrictions on the use of their creations. The fear of many AI experts is, in fact, that increasingly sophisticated systems could, in the future, automate critical decisions without human supervision.
A new frontier of power
If for years there has been talk of cyber war, now the confrontation moves to a further level: algorithmic warfare. Not a conflict made only of drones and missiles, but of mathematical models, simulations, probabilistic predictions. And in this scenario, whoever has the most advanced AI has a huge information advantage. The Claude case shows that technology brings with it worldviews, rules, political choices. And when these choices collide with the strategies of a state, the result is an open clash.
Because Claude is down today
Today, as global attention on generative AI continued to grow, Anthropic’s Claude platform has experienced difficult times: a sudden outage made login and use inaccessible to thousands of users around the world, precisely at the moment of maximum media momentum.
According to what was communicated from the companythe outage specifically affected Claude.ai and account sign-in and sign-out paths. Claude Code — the environment dedicated to developers — also experienced failures, while Claude’s API remained operational, allowing professional integrations to continue working without interruption.
As for the technical causes of the failure, Anthropic remains confidential for now: it has limited itself to declaring that it has identified the origin of the problem and is working to implement a definitive solution. However, several sources agree on one key piece of context: the incident occurred at a time of exceptional growth for the platform.
In fact, in recent days, Claude has rapidly climbed the App Store rankings, surpassing ChatGPT among the most downloaded apps. A success also fueled by the strong media exposure linked to the complex negotiations between Anthropic and the Pentagon, which brought the name of the chatbot to the center of the public debate on security, defense and artificial intelligence.
In the absence of a detailed technical explanation, the picture that emerges is that of an infrastructure put to the test by a sudden peak in traffic and new accesses. A paradox now familiar in the age of AI: popularity can itself become a vulnerability.
This story is about the future
The story is important, because it’s not just about the United States or Iran. It’s a global signal. AI is becoming an infrastructure of power, like energy or telecommunications. And, the real issue is not whether AI will be used in the military fieldthis has already happened, but to what extent will it be left free to operate and under what supervision. And above all, who will decide the limits of what an intelligent machine can do, when entering a field as delicate as national security.
