US Eases Export Ban on Anthropic's Mythos 5 AI Model

The US Commerce Department has partially lifted its export ban on Anthropic's Mythos 5 AI model, restoring access to more than 100 vetted US companies and agencies. This decision comes just two weeks after the department ordered Anthropic to block all foreign nationals from accessing both Mythos 5 and Fable 5 due to national security concerns. The models were deemed to have advanced cybersecurity capabilities that could be exploited by adversaries.

On June 12 and 13, the Commerce Department mandated that Anthropic restrict access to both models for non-US users. Anthropic complied swiftly, taking both models offline globally. However, on June 26, the government made an exception for Mythos 5, allowing vetted organizations, primarily those involved in defending critical infrastructure, to regain access. In contrast, Fable 5 remains completely restricted with no timeline for reinstatement.

This situation represents one of the first instances of the US government applying traditional export controls to advanced AI models. The initial ban was prompted by concerns over specific jailbreak techniques that could enable offensive cyber capabilities, categorizing these models as dual-use technologies subject to stringent licensing regulations.

Mythos 5 is specifically designed for cybersecurity applications, and the decision to keep Fable 5 fully restricted while reopening access to Mythos 5 suggests differing risk profiles for the two models. The partial lift of the ban creates a new dynamic in the AI market, as over 100 entities now have a functioning approval pipeline. This episode illustrates that frontier AI companies face risks similar to those encountered by semiconductor firms: products can be removed from the market overnight if deemed a national security issue.

The vetting process effectively grants a government endorsement to approved organizations, providing them with a competitive edge over others that do not receive such approval. As AI-powered security tools become increasingly vital for protecting decentralized finance (DeFi) protocols, auditing smart contracts, and detecting on-chain exploits, the implications of these export controls extend to any blockchain project relying on these advanced cybersecurity tools for their infrastructure.