From PGP to Mythos: a brief history of export controls that didn't stop anyone (techcrunch.com)

The White House has mandated Anthropic to restrict the export of its advanced AI models, Fable and Mythos, citing national security concerns. Following this directive, Anthropic quickly halted access to both models, marking the government's first substantial attempt to regulate frontier AI like it has with technologies such as encryption in the past. This episode raises significant questions about the effectiveness of U.S. export controls on powerful AI, especially as Anthropic had previously marketed Mythos as having potentially devastating capabilities if misused. The ban was prompted by concerns over security risk after a South Korean telecom's involvement with Mythos and a reported vulnerability in Fable 5 that Amazon researchers highlighted. This situation mirrors previous government failures in managing the proliferation of sensitive technologies and could lead to more stringent regulatory frameworks affecting AI labs nationwide. The outcome may reshape how American companies engage with international markets while spotlighting the challenges of enforcing compliance and safeguarding against dual-use technologies that could fall into the wrong hands. As history suggests, governments often struggle to control the spread of powerful software, and this latest incident may further complicate the landscape for AI development and export.