Anthropic has announced it will restore access to its Claude Fable model, one of the two frontier AI models whose worldwide availability was suspended on 13 June following a United States export control directive tied to national security concerns. The suspension, which also affected the more powerful Mythos model, cut off access for users outside the United States without prior warning, leaving Australian and other international customers unable to use services they had been relying upon and in many cases paying for, and generating significant frustration among individuals and businesses whose workflows had come to depend on the models. The restoration of Fable access represents a partial return to normal service, though the situation surrounding the more capable Mythos model, which was found by Britain’s AI Security Institute to be able to break into computer systems approximately 73 per cent of the time during testing, appears to remain unresolved as the more significant national security concerns are associated with that model’s capabilities rather than Fable’s.
The decision to restore Fable while the situation surrounding Mythos remains more complicated suggests that Anthropic and US authorities have drawn a meaningful distinction between the capability levels of the two models and the corresponding risk profiles they present when accessed internationally, with Fable sitting below the threshold of concern that triggered and continues to justify the more stringent controls applied to Mythos.