Anthropic shuts down Fable, Mythos models following Trump admin. directive

In a surprising turn of events, Anthropic has ceased access to its recently launched AI models, Mythos 5 and Fable 5, just days after their debut. This decision was triggered by a directive from the U.S. Commerce Department, which imposed export controls on these models, restricting their use beyond American borders. In an announcement made late Friday, Anthropic explained that compliance with the government order necessitated the immediate shutdown of both models for all users. The company reassured customers that access to its other AI models remains unaffected. Reports indicate that the administration's concerns stem from a jailbreak exploit that could potentially bypass safety measures designed to prevent Fable 5 from handling sensitive topics such as cybersecurity, chemistry, and biology. An official confirmed that the administration sought to delay the models’ release, allowing time for national security protocols to be strengthened against emerging threats. According to sources, this enhancement process could be finalized within weeks. In their communication, Anthropic highlighted that the government had only shared verbal evidence of a specific jailbreak that enables Fable 5 to analyze particular codebases for software weaknesses. The company noted that they had only observed this jailbreak being used to identify minor and relatively straightforward vulnerabilities, suggesting that other publicly available models, such as GPT-5.5, possess similar capabilities.