Amazon CEO Andy Jassy told senior Trump administration officials that his own researchers had pulled cyberattack-useful information from Anthropic’s flagship Fable 5 model. That warning triggered an export-control directive on Friday, June 12, 2026 that disabled Fable 5 and the related Mythos 5 for every customer worldwide. Anthropic disabled access globally, including for its own foreign-national staff, to comply.
The directive landed days after Anthropic publicly released Fable 5 with cybersecurity safeguards. In a blog post on Friday, Anthropic called the government’s disclosed jailbreak “narrow” and said other publicly available models, including OpenAI’s GPT-5.5, can identify the same minor flaws. “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” Anthropic wrote. The Information first reported Jassy’s outreach to the administration on Saturday, June 13.
The Directive Landed at 5:21 p.m.
Anthropic received the export-control directive from the U.S. government at 5:21 p.m. ET on Friday, the company said in a statement on its website. The letter, the company added, did not provide specific details of its national security concern. Citing national security authorities, the directive ordered Anthropic to suspend access to Fable 5 and Mythos 5.
The directive covers:
- Any foreign national, whether inside or outside the United States
- Including foreign national Anthropic employees
- All Anthropic customers worldwide, who now lose access to the models
The net effect of the order, the company wrote, is that it must abruptly disable Fable 5 and Mythos 5 for all of its customers. Access to all other Anthropic models, including Claude Opus 4.8, will not be affected, the company said.
Anthropic’s posture with respect to Fable’s safeguards, laid out in its launch blog post, included strong guardrails that “greatly reduce the likelihood that Fable is misused for tasks related to cybersecurity (among others).” In the weeks leading up to the launch, Anthropic said, it worked with the U.S. government, the UK AI Safety Institute, multiple private third-party organizations, and internal teams to red-team Fable’s safeguards for thousands of hours in total. Anthropic also requires 30-day retention of customer data with Fable, a policy change it said “carries real costs for us with customers, but that allows us to research and mitigate jailbreaks.” Anthropic has not received a disclosure of a concerning non-universal potential jailbreak that led to a harmful result, the company added.
Amazon’s Researchers Made the Call
The directive was prompted by conversations between Jassy and U.S. officials including Treasury Secretary Scott Bessent, people familiar with the matter told the Wall Street Journal on Saturday. Jassy told the officials that researchers at Amazon had used a series of prompts to get Fable 5 to provide them with information that could be used to aid cyberattacks and was supposed to be off-limits, the people said. Tech industry executives have been in regular touch with the administration about the power of cutting-edge AI tools, the Journal reported.
Amazon did not confirm whether it spoke to government officials about Anthropic’s models. “As a leading cloud provider that serves a large number of private and public sector customers, it’s not uncommon for governments to seek our counsel on potential security risks,” an Amazon spokesperson told Reuters.
“When they occur, we don’t share the details of these discussions,” the Amazon spokesperson added. The scoop on Jassy’s call to the administration was first published on Saturday, June 13, in a piece by AI reporter Stephanie Palazzolo. The Wall Street Journal confirmed the Treasury Secretary detail the same day, while Reuters reported on the broader directive.
The Information, citing a U.S. official, later reported that the administration is unlikely to force other AI firms to abide by restrictions similar to those placed on Anthropic. Reuters said it could not immediately verify that assessment. The U.S. Commerce Department’s Bureau of Industry and Security, which oversees export controls, did not respond to a request for comment on Saturday. The White House also did not respond to a request for comment, per WIRED. Anthropic said in its Friday blog post that it is “working to restore access as soon as possible.”
Anthropic Calls the Jailbreak Narrow
In its Friday blog post, Anthropic said the government’s concern centers on a narrow, non-universal jailbreak that essentially consists of asking the model to read a specific codebase and fix any software flaws. The company wrote that it reviewed a demonstration of the technique being used to identify a small number of previously known, minor vulnerabilities, and that those flaws appear relatively simple. No testers have yet been able to find a universal jailbreak on Fable’s safeguards, Anthropic added, though it conceded such jailbreaks will “eventually” be found in the future.
Other publicly available models can discover the same flaws without requiring a bypass, Anthropic added. “We have reviewed a report that we believe is the basis of the government’s directive and validated that the level of capability displayed there is widely available from other models (including OpenAI’s GPT-5.5), and is used every day by the defenders who keep systems safe,” the company wrote.
We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.
Anthropic also said in the post that the order did not include a disclosure of a non-universal potential jailbreak that led to a harmful result. The potential jailbreaks disclosed to the company are “either entirely benign responses or are minor findings that provide no Mythos-specific uplift,” the company wrote. The disclosure Anthropic reviewed is a single report, the company added, which it validated against other publicly available models including OpenAI’s GPT-5.5. Anthropic framed the directive as a misunderstanding and said it is “working to restore access as soon as possible.” If the disclosed standard were applied industry-wide, Anthropic warned, “it would essentially halt all new model deployments for all frontier model providers.”
A Ban That Reaches Anthropic’s Own Staff
The directive’s reach surprised even experts who favor export controls on advanced AI models. “This was not well thought-out,” said Jimmy Goodrich, a senior fellow at the University of California’s Institute for Global Conflict and Cooperation. “It even bans Canadian and British employees at Anthropic from doing research and development.”
Goodrich’s objection points to a structural feature of the directive that goes beyond the typical U.S.-China frame of AI export controls. By blocking any foreign national, inside or outside the U.S., from accessing the models, the order effectively forces Anthropic to wall off portions of its own workforce from its most advanced systems. Anthropic said in its blog post that the directive applies to “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.” The U.S. Commerce Department’s Bureau of Industry and Security, which oversees export controls, did not immediately respond to a request for comment. Red-teaming of Fable 5 had included the UK AI Safety Institute and other allied-government partners, the company said.
The Pentagon Feud That Set the Stage
The export-control directive is the latest flashpoint in a feud between Anthropic and the Trump administration that stretches back months. Earlier this year, the Department of Defense labeled Anthropic a “supply chain risk” after the Claude-maker sought to draw red lines over how the U.S. military could use its technology. The designation effectively barred government agencies and contractors from using Anthropic’s technology, and Anthropic responded by filing federal lawsuits against the Trump administration.
The public release of Anthropic’s first public Mythos-class model on Tuesday of that week was the company’s most ambitious rollout since the Pentagon dispute began. The Fable 5 launch was conducted in collaboration with the U.S. government, Anthropic said. The Mythos-class model had previously been tested under a limited rollout in April, the company said. Anthropic’s compliance with the supply chain risk designation, which Anthropic is challenging in court, came as the company was rolling out its most ambitious model yet.
Anthropic CEO Dario Amodei has been the public face of the company’s pushback against the Pentagon designation. The company has published its account of the supply chain risk designation, which the Department of Defense confirmed in a letter earlier this year.
The Quiet IPO Clock
The directive lands as Anthropic is preparing to go public. The San Francisco-based AI startup has confidentially filed for an IPO, Reuters reported. The order came as a separate dispute between the Trump administration and Anthropic had shown signs of easing across parts of the U.S. government. For an AI company with an imminent offering, a forced global shutdown of its flagship model is the kind of event that lands in regulatory filings and complicates enterprise sales pipelines. Anthropic apologized to customers in its Friday post and said it is “working to restore access as soon as possible.”
The company has committed to sharing more details about the government’s directive within 24 hours of receiving it. Access to Fable 5 and Mythos 5 has been disabled for all customers, including those in allied countries, Anthropic said. The directive lands at a delicate moment for Anthropic, which is preparing to go public and faces fresh uncertainty about its flagship model.
Amodei’s Pre-Mortem, Now Stress-Tested
Days before the directive, Anthropic CEO Dario Amodei published a policy essay laying out exactly the kind of process he said AI safety regulation should follow. Amodei’s framework for AI safety regulation argues that regulatory processes need to move faster than the technology itself.
The company invoked that framework in its Friday blog post, writing that the directive “does not adhere to those principles.” Anthropic argued the government should have “the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts.” The directive’s letter, which did not provide specific details of its national security concern, sits awkwardly with that framework. Amodei’s essay proposed concrete mechanisms for AI safety regulation that the company is now demanding the U.S. government adopt.
The Information, citing a U.S. official, reported that the administration is unlikely to force other AI firms to abide by restrictions similar to those placed on Anthropic. Reuters said it could not immediately verify that assessment. The U.S. Commerce Department did not respond to requests for comment on the matter on Saturday, according to WIRED.
Frequently Asked Questions
What is Anthropic’s Fable 5 model?
Fable 5 is Anthropic’s public release of a model it had previously tested in limited release as Mythos Preview. Launched on Tuesday of the week the directive landed, Fable 5 came with safeguards meant to prevent it from answering questions about cybersecurity, biology, and chemistry.
Why did the US government order Anthropic to disable Fable 5 and Mythos 5?
The U.S. government told Anthropic it believes it has become aware of a method of bypassing, or “jailbreaking,” a safeguard against using Fable 5 to find cybersecurity holes. The directive came in the form of an export control. Anthropic says the disclosed jailbreak is narrow and that other models, including OpenAI’s GPT-5.5, can identify the same minor flaws.
What did Amazon’s CEO reportedly tell the Trump administration?
Amazon CEO Andy Jassy told senior administration officials, including Treasury Secretary Scott Bessent, that researchers at Amazon had used a series of prompts to get Fable 5 to provide them with information that could be used to aid cyberattacks and was supposed to be off-limits, according to people familiar with the matter.
How does this affect Anthropic’s IPO plans?
Anthropic has confidentially filed for a U.S. initial public offering. A forced global shutdown of its just-released flagship model adds a fresh customer-trust and compliance overhang to that offering. Anthropic has apologized to customers and said it is working to restore access as soon as possible.
