Anthropic blacklisted by Pentagon, but NSA cleared to use Claude

American state has found itself in a paradoxical situation. The Pentagon officially blacklisted Anthropic as a national security threat to the supply chain, but intelligence agencies and the NSA are authorized to continue using the company's advanced models. The reason is simple: the state has no working alternatives.

How the Problem Arose

Anthropıc came under fire in 2025 for refusing to work with defense contractors. This was ethically sound from an AI ethics perspective, but created a serious dilemma for the American state. After banning chip exports to China and restricting access to US computing resources, the US faced an acute shortage of reliable software for deep data analytics. Anthropic controls a significant portion of this market through its flagship Claude model. The Pentagon officially added the company to the sanctions list, declaring it a threat to national security supply chains. On paper, this looked like an attempt at pressure, but in practice created unforeseen chaos.

Exemption as Forced Necessity

The intelligence community could not simply abandon the best tool on the market. The NSA and related agencies continued using Claude for critical data work, and no blacklist changed reality. As a result, the White House was forced to authorize a special exemption. National Security Advisor Susie Wiles approved the continuation of Anthropic's advanced model use by the NSA despite the company's official status. This is a rare and clearly uncomfortable decision that shows true priorities: when national security is at stake, pragmatism beats politics. Government sources unofficially acknowledged the substance of the problem:

• There is no real alternative to Claude at the frontier-model level on the market
• The ban damages intelligence more than hypothetical threat from Anthropic
• Geopolitical pressure only works if a working replacement exists
• Claude is embedded in critical analytics systems
• Exit would require years of retraining and tool rewriting

Lesson for the AI Industry

This paradox demonstrates a fundamental law of technology geopolitics. Ideological bans on companies only work in conditions of competition and availability of alternatives. If a competitor monopolizes access to the best tool, sanctions become merely a beautiful gesture at the diplomatic level. For Anthropic, this paradoxically works in its favor. Refusal of defense contracts was perceived as an ethics position, but in fact created a monopoly in practice. The state is officially against the company, but unofficially depends on it — this is more powerful than any contract.

What This Means

In the long term, this undermines stability. The US is investing in developing its own models, but in 2026 there are no results yet. As long as Claude remains the standard in intelligence, any complications with Anthropic will create a national security threat by the US's own hand.