Commerce Department restores Mythos 5 access for over 100 vetted entities while keeping Fable 5 completely locked down
The US Commerce Department reversed course on one of its most aggressive AI export restrictions, partially lifting a ban on Anthropic’s Mythos 5 model just two weeks after ordering the company to pull it offline entirely. The move restores access to more than 100 vetted US companies, agencies, and trusted partners, while a second model, Fable 5, remains completely restricted.
What happened and why it matters #
Between June 12 and 13, the Commerce Department instructed Anthropic to block all foreign nationals from accessing both Mythos 5 and Fable 5. National security concerns drove the decision, specifically around the models’ advanced cybersecurity capabilities and potential vulnerabilities that could be exploited by adversaries.
Anthropic complied within hours. Both models went dark for non-US users globally.
Then on June 26, the government carved out an exception for Mythos 5. Over 100 vetted organizations, primarily those defending critical infrastructure, regained access. Fable 5, however, stays completely offline with no timeline for reinstatement.
This is one of the earliest instances of the US government applying traditional export controls to frontier AI models. Concerns over specific jailbreak techniques potentially enabling powerful offensive cyber capabilities prompted the original intervention, categorizing the models as dual-use technologies subject to stringent licensing regulations.
The dual-use dilemma #
Mythos 5 is described as a cybersecurity-focused model. The decision to keep Fable 5 completely restricted while partially reopening Mythos 5 suggests the government views these two models as carrying different risk profiles.
What this means for investors #
The partial lift creates an interesting dynamic for the broader AI market. Over 100 entities getting restored access means there’s a functioning approval pipeline. The entire episode demonstrates that frontier AI companies now face a risk that semiconductor firms have dealt with for years: your product can be pulled from the market overnight if the government decides it’s a national security issue.
The vetting process creates a de facto government endorsement for approved organizations. Companies that make the list gain a competitive advantage, essentially receiving a federal stamp of approval that their competitors lack.
AI-powered security tools are increasingly central to protecting DeFi protocols, auditing smart contracts, and detecting on-chain exploits in real time. If the best AI models for cybersecurity are now subject to export controls and government vetting, that has downstream implications for any blockchain project relying on these tools for security infrastructure.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our