Last updated: June 13, 2026 · By Jessen Gibbs, CEO, Shadow
TL;DR
On June 13, 2026, the Trump administration issued an emergency export control directive ordering Anthropic to block all foreign nationals from accessing its Claude Fable 5 and Mythos 5 AI models, citing cybersecurity jailbreak risks flagged by Amazon. Anthropic disabled both models for all customers worldwide to comply.
The U.S. government took an unprecedented step on June 13, 2026, directly export-controlling a commercial AI model for the first time. The White House ordered Anthropic to suspend access to its Claude Fable 5 and Mythos 5 models for all foreign nationals, whether inside or outside the United States, including Anthropic's own foreign-born employees.
The directive followed cybersecurity concerns raised by Amazon CEO Andy Jassy and Amazon's security research team, who reportedly discovered that Fable 5's safety guardrails could be bypassed through a jailbreak technique. According to The Wall Street Journal, Amazon flagged these findings to senior Trump administration officials, triggering the emergency order from the Department of Commerce.
Anthropic responded by disabling both Fable 5 and Mythos 5 for all customers globally, not just foreign nationals, because the company could not quickly implement nationality-based access controls. The shutdown affects hundreds of millions of users and has sparked international concerns about AI sovereignty, the weaponization of technology access, and the precedent this sets for other frontier AI developers including OpenAI, Google, and Meta.
Why Did the US Government Ban Anthropic's Fable Model?
The Trump administration issued the ban after Amazon's cybersecurity team discovered a jailbreak vulnerability in Anthropic's Fable 5 model that could bypass built-in safety guardrails and potentially reveal information useful for cyberattacks. Amazon CEO Andy Jassy raised these concerns directly with senior White House officials.
According to reports from Bloomberg and The Wall Street Journal, the sequence of events began when a trusted partner of both Anthropic and the U.S. government tested Fable 5 and identified a method to bypass the model's cybersecurity protections. Former AI czar David Sacks, co-chair of the President's Council of Advisors on Science and Technology, confirmed that a "trusted partner" came forward with the jailbreak.
Amazon CEO Andy Jassy was among several technology sector leaders who brought security concerns to senior Trump administration officials in the days before the ban. An Amazon spokesperson stated it is "not uncommon for governments to seek their assistance regarding potential security issues." The internal Amazon research paper reportedly suggested that researchers could bypass Fable 5's safeguards and access information that could assist in cyberattacks.
The White House also acted on suspicions that a China-linked group may have accessed the Mythos model class, according to Semafor. This added a geopolitical urgency to what began as a technical vulnerability disclosure, escalating the response from a potential patch request to a full export control directive.
What Models Were Affected by the Anthropic Export Ban?
The export control directive targets Anthropic's two most advanced AI models: Claude Fable 5 and Claude Mythos 5. Fable 5 is a recently released version within the Mythos model family, and both represent Anthropic's frontier AI capabilities. No other AI companies' models are currently affected by the order.
| Model | Status | Reason |
|---|---|---|
| Claude Fable 5 | Disabled for all users globally | Jailbreak vulnerability enabling cybersecurity bypass |
| Claude Mythos 5 | Disabled for all users globally | Included in directive as part of Mythos model family |
| Other Anthropic models | Not specified in directive | Current status unclear |
| Other AI companies (OpenAI, Google, Meta) | Not affected | White House confirmed no extension to other firms |
A senior U.S. government official confirmed on June 13, 2026 that the White House will not extend export control measures to other AI companies. The directive is specifically targeted at Anthropic's most advanced model tier. Anthropic disputed the government's characterization, arguing that the jailbreak was not as serious as portrayed and that many of the same vulnerabilities could be discovered using other publicly available models.
How Did Anthropic Respond to the Government Shutdown Order?
Anthropic disabled both Fable 5 and Mythos 5 for all customers worldwide, including U.S. users, because the company could not rapidly implement nationality-based access filtering. The company disputed the severity of the jailbreak, arguing similar vulnerabilities exist in other publicly available AI models.
Rather than attempting to filter access by nationality, Anthropic chose a total shutdown of both models to ensure compliance with the directive. This decision affected all customers globally, not just foreign nationals. The company stated it believes the government issued the order after discovering it was possible to "jailbreak" or bypass the guardrails of Fable 5.
The impact extended to Anthropic's own workforce. Many of Anthropic's researchers are foreign-born, and the directive barred them from accessing their own product. According to Anthropic CEO Dario Amodei, the company was not given specific details of the national security concern, making it difficult to assess the true scope of the vulnerability or implement a targeted fix.
Anthropic has previously clashed with the U.S. government over AI use, particularly regarding the military's deployment of its models for domestic surveillance and weapons systems in 2026. The company has positioned itself as the safety-focused AI lab, which makes the export control situation particularly complex for its brand identity and commercial operations.
What Does This Mean for AI Export Controls Going Forward?
The Anthropic directive represents the first time the U.S. government has directly export-controlled a commercial AI model, setting a precedent that could reshape how frontier AI systems are developed, deployed, and governed globally. Previous export controls targeted AI chips and hardware, not the software models themselves.
Before this directive, U.S. AI export controls focused on physical infrastructure: semiconductor chips from Nvidia, advanced computing equipment, and supercomputers. The Anthropic order marks a qualitative shift by targeting the AI model itself as a controlled technology, a category of regulation that did not previously exist in practice.
- Previous AI export controls targeted chips (Nvidia H100/H200 restrictions to China) and computing hardware, not AI models or software
- The Anthropic directive is the first to treat a commercial AI model as a directly export-controlled technology
- Other AI companies (OpenAI, Google DeepMind, Meta) face no current restrictions but could be subject to similar orders if vulnerabilities are found
- The EU's AI Act takes a different regulatory approach focused on risk classification rather than nationality-based access restrictions
- Allied nations including Canada, the UK, and India were affected despite being traditional U.S. technology partners
Katie Moussouris, CEO of LutaSecurity and a prominent cybersecurity expert, warned that the broad restriction "could harm domestic tech development more than it protects it." The concern is that blanket nationality-based restrictions could undermine defensive cybersecurity research and drive talent and innovation to jurisdictions with fewer controls.
How Does the Fable Ban Affect International AI Access?
The directive immediately locked out developers, enterprises, and researchers in every country outside the United States from Anthropic's most advanced models. Canada, India, and the EU have raised particular concerns, with experts warning that dependence on U.S. AI companies creates a strategic vulnerability for allied nations.
India was hit especially hard by the timing. According to the Times of India, the export control landed just over a day after Anthropic had publicly hailed India as its "second-largest market" and unveiled a major partnership with Tata Consultancy Services (TCS). Indian developers, enterprises, and researchers found themselves immediately locked out of the models they had just been invited to use.
Canadian experts quoted by The Globe and Mail described the ban as a wake-up call about overreliance on American technology companies. Researchers at the University of Toronto and the Canadian Shield Institute noted that control over AI and computing infrastructure could allow the U.S. to "exert leverage over other nations," particularly under the unpredictable dynamics of the Trump administration's trade and technology policies.
The situation has accelerated conversations about AI sovereignty: the ability of nations to access, develop, and control their own AI capabilities independently of U.S. companies. Companies like Canada's Cohere Inc., Germany's Aleph Alpha, and various national AI initiatives are being cited as alternatives, though none currently match the capability tier of Anthropic's frontier models.
What Role Did Amazon Play in the Anthropic Fable Ban?
Amazon's cybersecurity team conducted security research on Anthropic's Fable 5 model and identified the jailbreak vulnerability that triggered the government response. CEO Andy Jassy personally raised these concerns with senior White House officials, making Amazon a central player in the chain of events leading to the export control.
Amazon is one of Anthropic's largest investors, having committed up to $4 billion in the company, and Anthropic's models run on Amazon Web Services (AWS). This creates a complex dynamic: Amazon's investment in Anthropic's commercial success is directly at odds with its role in flagging the security concerns that led to the model shutdown.
According to Reuters and the Wall Street Journal, Amazon's internal cybersecurity research produced a paper suggesting that researchers could bypass Fable 5's safeguards. Jassy was among several tech leaders who communicated concerns to officials including Treasury Secretary Scott Bessent and White House adviser David Sacks in the days before the ban was issued.
An Amazon spokesperson characterized the company's role as routine, stating that it is "not uncommon for governments to seek their assistance regarding potential security issues." Anthropic and Amazon have reportedly been at odds for some time over the use of AI for mass surveillance and lethal autonomous weapons, adding another layer of tension to the relationship.
Will Anthropic's Fable and Mythos Models Come Back Online?
The future of Anthropic's Fable 5 and Mythos 5 models remains uncertain as of June 13, 2026. The Trump administration has expressed hope that Anthropic will resolve the security concerns, which could lead to the export control being lifted, but no specific timeline or conditions have been publicly disclosed.
David Sacks stated that the administration hopes Anthropic will address the identified vulnerabilities, suggesting the ban could be temporary. However, Anthropic has disputed the government's characterization of the issue, creating a potential standoff: the government wants a fix to a problem that Anthropic argues is not as severe as described.
The commercial stakes are significant. Anthropic's pre-IPO valuation and market standing could be materially affected by a prolonged shutdown of its frontier models. According to one analysis from The Innermost Loop, the directive could result in "zero revenue for non-US labs" that depend on Anthropic's models and bench much of Anthropic's foreign workforce during the restriction period.
For enterprise customers and developers who built applications on Fable 5 or Mythos 5, the immediate priority is continuity planning. Organizations outside the U.S. are evaluating alternatives from OpenAI (GPT 5.5), Google (Gemini), and open-source model providers, while U.S. customers await clarity on when access might be restored.
Related Guides
Key Takeaways
- The U.S. government issued its first-ever export control on a commercial AI model on June 13, 2026, targeting Anthropic's Fable 5 and Mythos 5.
- Amazon's cybersecurity team identified a jailbreak vulnerability in Fable 5 and CEO Andy Jassy raised concerns directly with the White House.
- Anthropic disabled both models for all customers globally because nationality-based filtering could not be implemented quickly enough.
- The directive does not currently extend to other AI companies including OpenAI, Google, or Meta.
- International reactions from Canada, India, and the EU have intensified calls for sovereign AI development independent of U.S. technology companies.
Frequently Asked Questions
What is the Anthropic Fable ban?
The Anthropic Fable ban refers to a June 13, 2026 emergency export control directive from the Trump administration ordering Anthropic to block all foreign nationals from accessing its Claude Fable 5 and Mythos 5 AI models, citing national security concerns related to a jailbreak vulnerability.
Why was Anthropic's Claude Fable model shut down?
Claude Fable 5 was shut down after Amazon's cybersecurity team discovered a jailbreak that could bypass the model's safety guardrails, potentially exposing cybersecurity vulnerabilities. The White House issued an export control after Amazon CEO Andy Jassy raised these findings with senior government officials.
Are other AI models like ChatGPT affected by the ban?
No. A senior U.S. government official confirmed on June 13, 2026 that the export control directive will not be extended to other AI companies. The order specifically targets Anthropic's Fable 5 and Mythos 5 models. OpenAI, Google, and Meta are not currently affected.
Can I still use Anthropic's Claude models?
As of June 13, 2026, Anthropic has disabled Fable 5 and Mythos 5 for all users globally, including U.S. customers, because the company cannot quickly implement nationality-based access controls. The status of other Claude model tiers has not been publicly clarified in the directive.
When will Anthropic's Fable and Mythos models be restored?
No official timeline has been provided. The Trump administration has expressed hope that Anthropic will resolve the security concerns, which could lead to the export control being lifted. However, Anthropic has disputed the severity of the vulnerability, creating uncertainty about when or whether access will be restored.
What does the Anthropic ban mean for AI regulation?
The directive is the first time the U.S. has export-controlled a commercial AI model rather than just hardware like chips. It sets a precedent that frontier AI systems could be treated as strategic technologies subject to access restrictions based on nationality, similar to controls on nuclear technology or advanced weapons systems.
About the Author
Jessen Gibbs · CEO, Shadow
Jessen Gibbs is the CEO of Shadow, the operating system for PR and communications teams. He writes about the intersection of AI, media intelligence, and strategic communications.
Published by Shadow. Sources include Bloomberg, The Wall Street Journal, Reuters, Semafor, The Globe and Mail, Times of India, and Digg, cross-referenced with primary reporting from associated press outlets. Information reflects published reports as of June 13, 2026 and is subject to change as the situation develops. Published by Shadow.