US government orders Anthropic to block AI access

Anthropic has suspended global access to its advanced Claude Fable 5 and Mythos 5 AI models following a US government directive citing national security concerns. This abrupt action, occurring just days after the models' public release, impacts all users, including enterprises and employees, rerouting them to older versions. The company believes this is a misunderstanding and is working to restore access. This event highlights the vulnerability of cloud-based AI models to government oversight and serves as a critical warning for enterprises relying on single AI providers. The directive follows a reported jailbreak of the model, though Anthropic disputes the severity and uniqueness of the exploit, noting similar capabilities exist in rival models like OpenAI's GPT-5.5. This situation underscores the need for enterprises to diversify their AI supply chains and consider local, open-weight models for operational resilience.