Anthropic has withdrawn its Claude Fable 5 and Mythos 5 AI models from public access following security and national-security concerns [1].

The move signals a growing tension between rapid AI deployment and government oversight. As these models integrate deeper into global infrastructure, the risk of security vulnerabilities could have systemic implications for data privacy and national defense.

Authorities and Anthropic said security vulnerabilities and jailbreak risks were the primary reasons for the withdrawal [1, 2, 3]. These risks involve the potential for users to bypass safety filters, which could allow the AI to generate harmful content or expose sensitive system information.

Reports regarding the nature of the withdrawal vary. Geeky Gadgets said the U.S. government issued a ban on Claude Fable 5 due to security concerns [2]. However, Anthropic said the U.S. order to withdraw the model was a misunderstanding [3].

The withdrawal affects a global user base, particularly in the United States and India [2, 3]. India is the second-largest user base for Claude after the U.S. [3]. This wide reach means the sudden removal of the models disrupts workflows for millions of developers and enterprise users who had integrated Fable 5 into their systems.

Anthropic has not provided a specific timeline for when the models might return to public access. The company continues to evaluate the reported vulnerabilities to ensure the models meet safety standards before a potential relaunch [1, 2].

Anthropic has withdrawn its Claude Fable 5 and Mythos 5 AI models from public access

The conflict between Anthropic and U.S. authorities highlights the precarious balance between AI innovation and regulatory compliance. The discrepancy between a reported government 'ban' and the company's claim of a 'misunderstanding' suggests a lack of clear communication channels between AI labs and national security agencies. As AI models become more powerful, the threshold for what constitutes a 'security risk' is lowering, potentially leading to more frequent and abrupt service interruptions.