US Government Mandates AI Access Restrictions

Artificial intelligence developer Anthropic has confirmed it has received a directive from the United States government to restrict access to its cutting-edge AI models, Fable 5 and Mythos 5, for all foreign nationals. The order, which cites national security concerns, was communicated to the company on Friday, June 13, 2026, at 5:21 PM (21:21 GMT).

In a public statement, Anthropic, known for its Claude chatbot, detailed that the government instruction requires the immediate suspension of access to these specific AI models for individuals who are not US citizens. This restriction applies broadly, affecting even foreign nationals currently residing or working within the United States, including those employed by Anthropic itself. The company indicated that due to the sudden nature of the order, it had to implement the access cut-off with very little advance notice.

The Capabilities of Anthropic's AI Models

The core technology behind Anthropic’s Mythos AI model is particularly notable for its proficiency in identifying software vulnerabilities. This capability has been instrumental for US government agencies and select corporate partners in strengthening their cybersecurity defenses by pinpointing and rectifying security flaws, some of which had remained undetected for decades. However, the potential for such advanced AI to be misused as a potent cyberweapon has been a persistent concern among experts and authorities.

The Fable 5 model, which was introduced to the public only recently, is built upon the foundational Mythos technology. While it shares a common technological base, its most sensitive cybersecurity and biotechnology functionalities are deliberately blocked in the publicly accessible version. In contrast, Mythos 5 represents the full, non-public version of the technology. Its continued use is intended to be restricted to government entities and a select group of corporate partners, primarily for the purpose of fortifying their respective systems against cyber threats.

Limited Information and Company's Stance

Anthropic has stated that the government's order provided only partial information regarding the specific national security concerns that prompted the directive. The letter did not offer a detailed explanation of the underlying security issues. Following the receipt of the order, Anthropic's experts reviewed an internal report that they believe likely triggered the government's action. Their assessment suggests the concern revolves around a limited capability within the AI to analyze specific program code and correct errors.

The company also highlighted that similar capabilities are present in AI models developed by other providers, such as GPT-5.5 from its competitor, OpenAI. Anthropic expressed its disagreement with the decision to block access to software that is used by hundreds of millions of individuals globally for this reason. The company emphasized that the safety measures integrated into Fable 5 have undergone extensive testing and validation.

Broader Context of AI Development and Safety

This development occurs against a backdrop of ongoing discussions within the AI community regarding the rapid advancement and potential risks associated with sophisticated artificial intelligence systems. Earlier in June, Anthropic itself had advocated for a coordinated effort among leading AI companies to consider a pause in the development of highly advanced AI systems. The company had previously articulated concerns that AI technology is progressing at such an accelerated pace that there is a tangible risk of human oversight being diminished or lost entirely.

In a blog post published earlier in the month, Anthropic suggested that as cutting-edge AI becomes increasingly efficient and capable of performing complex tasks at higher speeds, it would be beneficial for the global community to have the option to slow down or temporarily halt its development. This perspective underscores a broader industry debate about balancing innovation with responsible development and risk mitigation in the rapidly evolving field of artificial intelligence.

Source: Original Article