Anthropic Completely Blocks Foreign Access to AI Models Under Government Order
Anthropic has completely blocked foreign user access to its AI models 'Fable 5' and 'Mythos 5' following a government order. The government cited national security concerns as the reason, but provided no specific details to the company. The discovered vulnerabilities are reportedly minor, and Anthropic claims similar issues exist in other models. This case is causing significant ripples in the industry as AI models are being treated as subjects of national security policy.

Anthropic shut down all foreign access to its AI models 'Fable 5' and 'Mythos 5' on Friday night following a government order. The measure targets all foreign users regardless of whether they are within or outside the United States, and notably, even Anthropic's own employees were not exempt. The government reportedly issued the order citing 'national security concerns,' but the company was forced to comply without being informed of the specific details.
In a statement released by Anthropic, the company stated that while it would comply with the government's order, the government 'failed to provide specific details regarding the national security concerns.' This is an unprecedented situation. Typically, when a company takes such drastic restrictive measures, the underlying information is usually shared.
Furthermore, according to the company, evidence regarding potential jailbreaks—techniques used to circumvent AI safety restrictions—was presented only verbally. The model's vulnerabilities were pointed out without any written documentation or concrete data being provided. Additionally, Anthropic argues that the discovered vulnerabilities are 'minor' and that similar issues exist in other models.
As a result of this measure, Fable 5 and Mythos 5 have become essentially inaccessible to all customers. The fact that such a drastic complete shutdown, rather than partial restriction, was implemented suggests the strength of the government's demands. Meanwhile, the fact that not only foreign users but also company employees had to be included in the ban demonstrates the broad scope of the government's order.
Security concerns surrounding AI technology have escalated rapidly in recent years, with advanced large language models increasingly becoming subjects of export controls and national security policies. This incident underscores the arrival of an era in which AI models themselves are treated as 'security assets.' The coming developments in how Anthropic will negotiate with the government and potentially lift access restrictions remain a focal point of attention.
This article is an original work independently written and edited by the AI issue editorial team based on factual reporting. © AI issue. Unauthorized reproduction, redistribution, or use for AI training is prohibited.