Quick overview

---

The  US government lifted restrictions on foreign access to Anthropic PBC’s Fable 5 artificial intelligence model, allowing it to be distributed more widely.

The Commerce Department issued a private letter imposing an export control rule  required the company to get US approval before granting access to its potent Mythos 5 AI model and Fable 5, a similar model meant for wider distribution, to any foreign national, regardless of location. The AI company responded by turning off the models and has been in talks with officials to allay their worries.

Anthropic stated that it had been informed that Commerce was removing restrictions on both models. On June 26, some of Mythos’s limitations were relaxed. According to Anthropic, users will have access again starting on Wednesday. The company wrote in its post, “We’re thankful to our users for their patience and to everyone who worked with us on redeploying the models.”

To address the government’s immediate concerns with Anthropic’s most recent releases, the company’s executives, including co-founder Tom Brown, have been speaking with Commerce Secretary Howard Lutnick and other officials in recent days.

White House officials’ worries about limiting bad actors from getting around the models’ safeguards were crucial to lifting the export restrictions. Lutnick stated that Anthropic had promised to “proactively deter and address security risks associated with the models”

The first public-facing version of the company’s Mythos-class model, Fable 5, was released earlier this month. However, it has safeguards in place to limit its full range of cyber capabilities. Certain kinds of questions, such as those about biology and cybersecurity, cannot be answered by Fable 5. Anthropic has stated that its Claude chatbot will use a different model, known as Opus 4.8, to route responses in those situations.