How the conflict between Anthropic and the Pentagon can affect the field of AI. Analysis

The US government has demanded that Anthropic close access to the recently released versions of the Claude Mythos 5 and Claude Fable 5 artificial intelligence models to all users who are not US citizens, explaining this for national security reasons. At the same time, Anthropic continues to have a conflict with the Pentagon over the company's ban on using its products for military purposes. Why the US authorities are fighting the most effective AI developer and how this confrontation will affect the entire artificial intelligence industry — in the Izvestia article.

The official reason for the ban

• Access to Mythos 5 and Fable 5 was closed on the evening of June 12, three days after its official release to the market, due to an urgent report from Amazon, which, by the way, is a major investor in Anthropic. The details of the report were not disclosed, but based on the comments of Anthropic, we are talking about the fact that some functions of Fable 5 allow it to be used for cyber attacks, which may pose a threat to national security.

• Mythos 5 and Fable 5 are configurations of the improved Claude 5 base model, with the Mythos AI model designed for a limited range of corporate clients, and Fable is its custom version. While advertising Claude Mythos 5, Anthropic noted that it was delaying the release of the AI assistant due to the fact that it is too powerful a tool, and in particular, it is able to detect vulnerabilities in software faster and better than a human. But such statements have already become a sustainable marketing strategy when promoting new AI models to the market, including Gemini and Chat GPT.

• The United States demanded that access to Mythos 5 and Fable 5 be closed to users who do not have US citizenship, including employees of the company itself. Due to the inability to track the nationality of its customers, Anthropic was forced to block new products for all users. In its defense, the company claims that the AI model uses the simplest circumvention mechanism, and other AI models on the market have the same ability.

• It is noteworthy that both configurations of the Claude 5 base model had an unprecedented level of protection compared to other AI agents. In particular, Mythos 5 was developed in collaboration with the US government, and the user version of Fable 5 includes a ban on instructions on performing dangerous actions, access to dual-use information and "adult" content. In this regard, Anthropic is noticeably ahead of its competitors. But now the United States is actually imposing even stricter requirements on AI assistants, which none of the existing models can yet meet.

Ideological contradictions

• The issue of values has become a fundamental disagreement between the company and the US authorities. The founders of Anthropic left Open AI in 2020-2021 due to the fact that they put ethical issues of artificial intelligence in the first place. Dario Amodei, co-founder and CEO of Anthropic, outlined his views in the Machines of Loving Grace manifesto. Anthropic became the first public benefit company in the field of AI, with the goal of "developing and maintaining safe and responsible artificial intelligence."

Ethical force: Claude's AI, "cancelled" in the USA, retains leadership in downloads

The company Anthropic, which has fallen out of favor with Trump and the Pentagon, is becoming more popular because of its commitment to its principles

* The conflict between Anthropic and the Pentagon began after the company learned that its AI model was used to plan an operation to kidnap Venezuelan President Nicolas Maduro on January 3, 2026. The company insisted that the cooperation agreement stipulated a ban on the use of Anthropic products for military purposes. In response, the Pentagon demanded that the company lift the restriction, threatening to sever ties with Anthropic and assign it the status of a "supply chain risk", which is usually assigned to companies associated with "hostile" states and under sanctions.

• After the refusal of Anthropic to lift restrictions on the use of its products, US President Donald Trump called for an end to using the company's technologies, and on March 1, the United States launched strikes against Iran, and the Claude AI model developed by Anthropic again played a key role in the military operation. It was integrated into Palantir under the name Maven (about the ideology of Palantir, we it was told here). In March, the company received confirmation from the Pentagon that it would be assigned a "risk status" and appealed to a federal court, which blocked the Pentagon's "sanctions" against the company, but on April 8, the Court of Appeals for the District of Columbia recognized the government's actions as legitimate.

• The United States is putting pressure on the company at a time when Anthropic is preparing to launch an IPO, and the sub-sanctioned status may drop the company's capitalization and the starting value of its shares. The ban on access to the product also hurt the image of Anthropic and caused dissatisfaction among customers both in the United States and abroad: having already paid for access to a more powerful model, they are forced to use older versions of Claude.

Bit of bad luck: ChatGPT and Grok pointed out Trump's mistakes in planning the war in Iran

How the use of neural networks led to an ethical scandal between the Pentagon and IT giants

Implications for the AI industry

• In the European Union, the ban on access to new Anthropic products was perceived as an alarming signal: for the first time, the United States blocked the EU's access to advanced American technologies. The European Commission noted that this case "underscores the need for technological sovereignty for Europe," and the countries of the union called for increased investment in their own development of AI models. The draft law on the development of cloud technologies and artificial intelligence proposed by the European Commission on June 1 already involves restricting access to government contracts for American companies Amazon, Microsoft and Google.

Formally, the situation with Anthropic was the first case of strict regulation of the field of AI and blocking the spread of the latest technologies due to the risk of national security. The company itself advocated for control over the development and application of artificial intelligence models, especially since the world has already faced the negative effect of lack of control over this area, including the spread of deepfakes and phishing mailings. But the possibilities of regulation by the United States are limited: the advantage of individual companies in the field of AI is short-term, and more powerful models from Chinese developers may soon appear, which the United States will not be able to control.

• The AI industry is facing a new reality where the government can bring the most advanced AI model to market. Moreover, these actions seem to have been provoked by a marketing ploy that all companies use. In addition, blocking access to new products from Anthropic means that the company will not make a profit, and further development remains questionable. Investors are used to the fact that AI models are constantly being improved, and if their development is stopped, then interest in companies going public (and Open AI has such plans) will decrease, and the fall in shares of AI companies may lead to the "bubble" in the AI market bursting.