TL;DR
White House AI adviser David Sacks said Anthropic’s Fable models were restricted after the company refused to fix a jailbreak tied to cyber capabilities. Anthropic disputes that account, saying officials gave no technical detail and that the flaws were minor and reproducible in other models.
White House AI adviser David Sacks said the U.S. government restricted Anthropic’s most powerful Fable models after the company refused to fix or withdraw a jailbreak that he described as restoring cyberweapon-level capability, a claim Anthropic rejects as unsupported and overstated.
Sacks, co-chair of the President’s Council of Advisors on Science and Technology, wrote on X on June 13 that a “highly credible trusted partner” found a bypass of Fable’s guardrails. According to his account, officials asked Anthropic CEO Dario Amodei to fix the issue or pull the model, and the company refused. Sacks said the export control was issued “reluctantly.”
Anthropic said in a June 12 blog post that the government did not provide specific technical detail. The company characterized the demonstration as showing a few minor, already-known weaknesses and said similar behavior could be reproduced in other public models, including GPT-5.5, without a special bypass.
The core dispute is severity. Sacks describes the issue as one that could restore the operability of a cyberweapon. Anthropic describes it as a narrow potential jailbreak that should not trigger a recall of a model used by hundreds of millions of people. No public technical record has been released that would allow outside researchers to judge which account is closer to the facts.
The Safety Card, Played From Every Side
● ContestedA White House adviser says Anthropic refused to fix a cyberweapon jailbreak and got banned for it. Anthropic says the flaw is trivial. Almost every fact that would settle it is non-public — and “safety” is now the card every side is playing.
Both are claims, not findings. They don’t disagree on tone — they disagree on what the bypass actually is.
- A “highly credible trusted partner” found a jailbreak of Fable’s guardrails.
- The admin asked Amodei to fix it or pull the model. He refused.
- So the export control was issued — “reluctantly.”
- It restores operability of a cyberweapon; calling that “not serious” is indefensible.
- The government gave no specific technical detail.
- The demo found a few minor, already-known flaws.
- Other public models (incl. GPT-5.5) do the same without a bypass.
- A “narrow potential jailbreak” shouldn’t recall a model used by hundreds of millions.
Per reporting by Semafor (carried by Fortune and others), the entity that flagged the jailbreak was Amazon — with CEO Andy Jassy reportedly in contact with the administration. Amazon hasn’t confirmed specifics. Flagging a real risk is what a good partner does — but Amazon wears three hats at once, and none of them is neutral.
Each actor’s safety claim points toward its own advantage.
The entire evidentiary record is a matter of trusting parties who each have a reason to shade it.
A transparent, technically grounded, independently reviewable process — which is, notably, exactly what Anthropic says it wants, and exactly what would also constrain Anthropic. The reason to demand it isn’t loyalty to anyone; it’s that the alternative is decisions made on secret evidence and adjudicated in dueling press statements.
Independent commentary, produced with AI assistance under human editorial oversight; the views are the author’s own and may change. This is analysis and opinion, not investment, financial, legal, or technical advice, and it concerns an actively developing situation in which key facts are disputed and non-public. Claims attributed to David Sacks reflect his June 13, 2026 statement on X; claims attributed to Anthropic reflect its published statements; reporting on Amazon’s role reflects accounts published by Semafor and others — all read as of June 15, 2026, and presented as the claims of those parties, not as established fact. Characterizations are the author’s interpretation, offered in good faith and open to rebuttal. References to specific people, companies, and government actions are factual and analytical, not partisan, and imply no affiliation or endorsement.
Secret Evidence Meets AI Controls
The dispute matters because it appears to involve one of Washington’s most forceful interventions yet in commercial AI deployment, based on evidence the public cannot inspect. If Sacks’ account is accurate, the government acted to prevent a high-capability model from being used in ways Anthropic’s own safety framing had warned against. If Anthropic’s account is accurate, a major AI model was restricted over a defect the company says was limited, familiar and not unique.
The case also shows how “safety” arguments now cut in several directions. Government officials can invoke safety to justify limits on deployment. AI companies can use safety claims to support regulation of rivals or defend their own models. Commercial partners can flag risks while also having business interests in the outcome.
Artificial Intelligence for Cybersecurity: Develop AI approaches to solve cybersecurity problems in your organization
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Anthropic’s Safety Argument Turns Back
The dispute centers on Fable and Mythos, as described in the source material. Sacks said Fable is effectively Mythos with guardrails, and that a guardrail failure would expose Mythos-level cyber capabilities. Anthropic has previously argued that advanced AI systems with cyber potential require regulation, making the current fight unusually pointed: the safety framework it supported is now being applied to its own product.
The identity of the “trusted partner” has not been confirmed by the government. Reporting by Semafor, carried by Fortune and others, said Amazon may have been the entity that flagged the jailbreak, with Amazon CEO Andy Jassy reportedly in contact with the administration. Amazon has not confirmed those specifics in the provided source material.
That possible role would carry added weight because Amazon has several relationships to the companies and market at issue: it has invested billions in Anthropic, supplies compute for Anthropic, and also competes in AI models. Those facts do not prove improper conduct, but they make transparency over the evidence and process more pressing.
“A highly credible trusted partner found a jailbreak of Fable’s guardrails.”
— David Sacks, White House AI adviser, on X
Modern Offensive Cybersecurity with Agentic AI: Leverage MCP, n8n, and AI Agents for Advanced Security Testing (The AI knowledge Library)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Almost all facts needed to resolve the dispute remain non-public. The government has not released the technical details of the alleged jailbreak, a published methodology, a CVE-style record, or an independent review. The trusted partner has not been named by officials, and Amazon’s reported role remains unconfirmed in the source material.
It is also not clear whether the issue is unique to Fable, whether it allows materially greater cyber capability than other public models, or whether Anthropic was given enough information to reproduce and patch the problem. The current public record is made up of competing claims from parties with stakes in the result.
AI jailbreak detection software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Patch Talks And Ban Timing
The next signal will be whether the restriction is lifted quickly after a quiet fix or remains in place. A fast reversal would suggest officials and Anthropic reached a technical remedy, though it would not by itself prove the original flaw was severe. A longer standoff would increase pressure for a reviewable process that lets independent experts test the government’s claim and Anthropic’s response.
For now, readers should treat the dispute as unresolved. The confirmed development is the public clash between Sacks’ account and Anthropic’s denial, not the underlying technical finding.
ChatGPT for Cybersecurity Cookbook: Learn practical generative AI recipes to supercharge your cybersecurity skills
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened to Anthropic’s Fable models?
David Sacks said the U.S. government restricted Anthropic’s most powerful Fable models after a trusted partner found a jailbreak and Anthropic refused to fix or withdraw the model. Anthropic disputes the severity and basis of that action.
Is the alleged Fable jailbreak confirmed?
A jailbreak has been alleged by Sacks and disputed by Anthropic. The technical details have not been made public, so outside verification is not possible from the current record.
What does Anthropic say about the government’s claim?
Anthropic says officials provided no specific technical detail and that the demonstration involved minor, already-known flaws that can be reproduced in other public models.
Why is Amazon part of the story?
Semafor reporting carried by other outlets said Amazon may have been the trusted partner that flagged the issue. That role is unconfirmed in the provided material, but it matters because Amazon is an Anthropic investor, cloud provider and AI competitor.
What would resolve the dispute?
A public or independently reviewable technical assessment would help determine whether the flaw was unique, severe and properly handled by the government and Anthropic.
Source: Thorsten Meyer AI
![GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]](https://deepintellica.com/wp-content/uploads/2026/07/gpt-5-6-sol-ultra-produces-proof-of-the-cycle-double-cover-conjecture-pdf-featured-1-260x140.jpg)
