“The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” Anthropic said in a statement Friday evening, News.Az reports, citing Politico.
***
Fable 5 is a less powerful version of the company’s Mythos model, which Anthropic initially withheld from a larger audience for fear that the model was too capable of exploiting software vulnerabilities.
The company claimed the U.S. government imposed export controls out of concerns that some of the models’ built-in safeguards could be bypassed, a process known as “jailbreaking.”
“To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak,” the company said. It added that it disagreed “that the finding of a narrow potential jailbreak should be cause for recalling a commercial model.”
In early April, Anthropic announced it had granted a limited set of trusted tech and cyber firms access to Mythos Preview through the Glasswing project to help them boost cyber defenses.
The decision left many non-U.S. governments and entities, such as the EU institutions, scrambling to get access.
Fable 5, which was launched only this week, is described as a “Mythos-class model” that the company provided with safeguards to make it safe for general use.
“Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage,” Anthropic said at launch.
Anthropic’s decision to pause access to its models sparked comments in Europe and the U.K. over the need to become more sovereign from U.S. AI models.
“As we debate the future of national security and technological sovereignty, access to AI capabilities is crucial,” said Kanishka Narayan, the U.K.’s minister for AI.
