The US company Anthropic has temporarily withdrawn its new AI model from the market due to security concerns raised by US authorities. The company announced these restrictions only a few days after the program’s public unveiling.
According to a statement on Anthropic’s website, it was instructed to bar foreign users from accessing Claude Fable 5. Anthropic itself had previously described the program as “too powerful.” The company clarified the immediate impact, stating, “The result of this order is that we must immediately shut down Fable 5 and Mythos 5 for all our customers in order to comply with the requirements.”
Claude Fable 5 is a version of the Claude Mythos program, competing in the market with services such as OpenAI’s ChatGPT and Google’s Gemini. Although Anthropic noted that US security agencies had not named specific dangers, the company indicated that the government allegedly learned of a method that could circumvent Fable 5’s safety features-a process known as “jailbreaking.”
Anthropic explained that it examined a demonstration of this method, which allowed them to identify a handful of minor, pre-existing vulnerabilities. These flaws were described as relatively simple, suggesting they could potentially be found in other publicly available AI models without needing an exploit bypass.
Anthropic has long been under scrutiny by the US government. President Donald Trump had publicly criticized the company, and Defense Secretary Pete Hegseth later classified it as a “supply chain risk”-a designation never before applied to a US company. This classification signifies that a product or service is deemed insufficiently secure for use by government agencies, a designation previously reserved for entities from hostile states. In response, Anthropic sued the Pentagon, and the order issued by the Pentagon has since been provisionally suspended.
