Anthropic blocked access to flagship models due to US pressure

Anthropic blocked access to its advanced AI models for all users without exception. The reason is a new escalation of tensions with US authorities, who have classified the company as a national security threat since spring of this year.

The dispute that has been dragging on since spring

In early 2025, US regulators added Anthropic to a list of organizations potentially posing a threat to US national security. This status entails enhanced oversight, restrictions on certain operations, and possible pressure to disclose information about technologies. The paradox of the situation is that Anthropic is arguably the most consistently security-focused AI laboratory among leading players.

Founded by OpenAI alumni, the company publicly supports government regulation of AI, invests in alignment research, and participates in developing international AI safety standards. It is precisely such a company that came under suspicion — which says much about the logic of the regulatory process. Since spring, Anthropic has been formally contesting the status assigned to it.

In recent months, interactions with authorities shifted to a more constructive mode: both sides demonstrated readiness for dialogue, and it seemed that tensions were gradually subsiding.

What happened this week

Something disrupted the balance this week. The details of the new conflict are not being disclosed, but the result is clear: Anthropic was forced to block access to advanced models for all users. The scale of the blockade is notable: this is not about restrictions in specific markets or for specific categories of clients — access is closed entirely. Apparently, restrictions affected the flagship models in the Claude line, including the most powerful versions. For corporate clients building products on Anthropic's API, the situation means immediate disruptions. Switching to alternatives — GPT or Gemini — requires time and resources, especially with deep integration.

• Since spring 2025, Anthropic has been contesting its US national security threat status
• In recent months, dialogue with authorities has been constructive
• A new conflict emerged this week — its nature has not been disclosed
• All users have lost access to advanced models
• Timelines for lifting restrictions have not been announced

Risks for the entire market

The Anthropic case exposes a structural risk for companies building infrastructure on third-party AI provider APIs. If access to flagship models can be blocked due to regulatory pressure, the reliability of an AI stack depends not only on the technical stability of the provider but also on its relationship with the government. For the market as a whole, this is a troubling precedent.

Anthropic is one of the companies most loyal to regulation: it openly supported AI oversight long before it became fashionable. If this is happening to them, other players — less inclined to dialogue with authorities — may face even harsher consequences. Against the backdrop of the global race for AI leadership, where the US seeks to maintain its advantage, regulatory control over leading laboratories will only intensify — regardless of the companies' own declarations.

What this means

The Anthropic case is a clear reminder that even the largest AI laboratories operate under conditions of growing regulatory uncertainty. For developers and corporate clients, the conclusion is unambiguous: building critical infrastructure on a single AI provider is becoming an increasingly risky strategy. Diversification, which a year ago seemed like excessive caution, today becomes a basic requirement for reliable architecture.