Anthropic, the AI company behind the Claude chatbot, has blocked access to its newly launched advanced software after receiving an order from the U.S. government. In a blog post on Friday (June 5, 2026), the company stated that government agencies demanded it prevent all foreign nationals, including those working at Anthropic in the U.S., from accessing the Fable 5 and Mythos 5 models due to national security concerns.
According to Anthropic, the order was received at 5:21 PM GMT on the same day, and the letter did not provide specific details about the security threat. As a result, the company was forced to cut off access for everyone in a short time frame.
The AI behind Anthropic's Mythos model can detect software vulnerabilities, including some that have existed for decades and remained undiscovered. This capability has been used by the U.S. government and select companies to patch security flaws. However, from the start, there were fears that such AI could become a dangerous cyber weapon if it fell into the wrong hands.
The Fable 5 model, released earlier this week, is based on Mythos technology but has had its cybersecurity and biotechnology capabilities disabled. Meanwhile, Mythos 5 is the full version, not publicly available, and is only used by government agencies and selected corporate partners to strengthen their systems.
Anthropic emphasized that it had only received partial information from the government and said it had reviewed a report that it believed may have prompted the order. The company's experts concluded that the report referred to limited capabilities in using AI to review specific code and fix bugs. Anthropic stressed that models from other providers, such as rival OpenAI's GPT-5.5, possess similar abilities.
Anthropic stated it disagrees with blocking software used by hundreds of millions of users for this reason and asserted that the safety measures in Fable 5 had been thoroughly tested. Earlier in early June, the company proposed that the world's leading AI companies coordinate a pause in the development of advanced AI systems, warning that the technology is improving too quickly and risked humans losing control.
