White House blocks Anthropic from exporting Fable and Mythos AI models

In short: The White House ordered Anthropic to restrict access to its Fable and Mythos AI models outside the United States, and Anthropic pulled both models offline.

What happened

The White House ordered AI company Anthropic to stop “exporting” two of its most powerful AI models, called Fable and Mythos. In this case, export means making the models available to people outside the United States, and even to some foreign nationals inside the country.

According to TechCrunch, Anthropic shut off access to both models shortly after getting the order. The models have been unavailable for about a week. The government cited national security concerns, but did not publicly give details.

TechCrunch reports two events may have contributed to the decision. One was Anthropic granting access to a South Korean telecom through a limited partner program, after US officials suspected the telecom had ties to China, which the company denied. Another was Amazon CEO Andy Jassy reportedly warning officials that Amazon researchers found a way around some protections in Fable 5, which Anthropic said was a narrow issue that it already fixed.

Why it matters

This is an early test of whether the US can use export controls to limit access to advanced AI, like trying to keep a powerful tool from being shipped overseas. The article argues that similar efforts in the past have struggled, including attempts to restrict encryption (a way to lock messages so only the right person can read them) and spyware (software used to secretly watch or break into devices). What happens next could affect whether US AI companies need government approval before serving customers in other countries.

Source: TechCrunch AI