Foreign national access to Fable and Mythos, Anthropic’s most advanced AI models, has been abruptly suspended worldwide following a directive from the US government. Citing national security concerns, the order mandates that Anthropic block access for all foreign nationals, including its own employees, effectively taking both Fable 5 and Mythos 5 offline for all users globally.
The directive, issued on June 12, 2026, at 5:21 PM ET, immediately halted the rollout of Fable 5, which had only begun reaching Pro, Max, and Enterprise customers three days prior. This sudden move impacts millions who were just gaining free access to the powerful AI model, designed for advanced applications.
US Government Bars Foreign National Access to Fable, Mythos
Anthropic stated it received the export control directive under “national security” authorities. The broad scope of the order means that Fable 5 and Mythos 5 are now unavailable, regardless of user location or nationality. All other Anthropic models, including Claude Opus 4.8, remain operational.
Fable 5 and Mythos 5 share a common underlying architecture. Fable 5 incorporates safeguards to filter sensitive queries related to cybersecurity, biology, and chemistry. In contrast, Mythos 5, the unrestricted version, was previously available only to vetted government cyberdefenders and life sciences partners. The government’s intervention now restricts both, highlighting the escalating regulatory scrutiny on frontier AI capabilities.
“To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws.”
In a developer notice, Anthropic advised integrators to migrate to other models, as new sessions would default to Opus 4.8 and existing Fable 5 sessions would terminate with errors. The UK’s Minister for AI and Online Safety, Kanishka Narayan MP, quickly weighed in, noting the impact on both US and UK customers and underscoring the broader implications for technological sovereignty, referencing the UK’s significant AI chip investment.
The Jailbreak Controversy and Industry Implications
Anthropic’s understanding is that the directive stems from a reported method to ‘jailbreak’ Fable 5. However, the company reviewed the demo and characterized the findings as minor, already-known bugs, comparable to vulnerabilities discoverable in other publicly available models without special bypasses. They argue that applying such a stringent standard to commercial models could stifle innovation across the entire AI industry.
The company maintains that the capability demonstrated in the alleged jailbreak is widely available in other models, such as OpenAI’s GPT-5.5, and is routinely utilized by cybersecurity defenders. This perspective suggests a potential misunderstanding on the government’s part regarding the practical applications and existing landscape of advanced AI models. The sudden restriction on foreign national access to Fable and Mythos has sparked debate over the balance between national security and technological advancement.
Anthropic’s Path Forward
Despite complying with the legal directive, Anthropic has expressed disagreement with the rationale, stating that a ‘narrow potential jailbreak’ should not warrant the recall of a commercial model deployed to hundreds of millions. The company is actively working to restore access and has promised to provide more details within 24 hours, indicating ongoing discussions with government officials to clarify the situation and potentially reinstate services.
This incident underscores the complex challenges at the intersection of advanced AI development, national security, and international collaboration. As AI models become more powerful and pervasive, governments worldwide are grappling with how to regulate their deployment and ensure responsible use, particularly concerning sensitive applications. The future of foreign national access to Fable and Mythos, and indeed other frontier AI models, remains uncertain as these discussions evolve.
