TL;DR
White House AI adviser David Sacks says Anthropic’s Fable models were restricted after the company refused to fix a jailbreak that could restore Mythos-level cyber capability. Anthropic says the government has not provided technical proof and describes the issue as narrow and minor. The dispute matters because a major AI safety action now rests on evidence the public cannot review.
White House AI adviser David Sacks says the U.S. government restricted Anthropic’s most powerful Fable models because the company refused to fix or withdraw a jailbreak that allegedly restored dangerous cyber capabilities, while Anthropic says the reported flaw is minor, already known and not grounds for a model recall affecting a large user base.
Sacks, co-chair of the President’s Council of Advisors on Science and Technology, said in a June 13, 2026, statement on X that a “highly credible trusted partner” found a bypass of Fable’s safety guardrails. According to Sacks, the administration asked Anthropic CEO Dario Amodei to fix the issue or pull the model, and the export-control action followed after Amodei refused.
Anthropic has disputed that account. In a June 12 blog post, the company said the government did not provide specific technical detail and that the demonstrated behavior reflected “a few minor, already-known flaws.” Anthropic also said comparable outputs could be reproduced on other public models, including GPT-5.5, without using a special bypass.
The central facts needed to settle the dispute remain outside public view. The government has not released a technical write-up, methodology, CVE-style disclosure, or independent review. Anthropic has not published a full reproduction record either, leaving readers to weigh competing claims from parties with direct stakes in the outcome.
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 Oversight
The case tests how far U.S. officials can go in limiting access to a commercial AI model based on safety findings that are not publicly reviewable. If Sacks’s account is accurate, the government acted to prevent a model from enabling cyber operations that Anthropic itself had previously warned should be regulated. If Anthropic’s account is accurate, the government may have taken an unusually forceful step over a narrow flaw that exists across the market.
The dispute also affects trust in AI safety governance. Companies, cloud providers and government officials all use safety claims in public arguments, but those claims point to different commercial and policy outcomes. Without a reviewable standard, model restrictions can appear to depend on private briefings and dueling statements rather than a shared technical record.
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.
How the Fable Ban Emerged
The dispute centers on Fable, described in the source material as a guarded version of Anthropic’s Mythos system. Sacks’s argument relies on the claim that if Fable’s guardrails fail, the system can return to Mythos-class cyber capability. Anthropic has previously argued that highly capable cyber models require stronger regulation, a position critics say can also protect incumbents by raising barriers for rivals.
Reporting by Semafor, carried by Fortune and other outlets according to the source material, said the “trusted partner” that flagged the jailbreak may have been Amazon. Amazon has not confirmed the specifics cited in that reporting. The alleged role matters because Amazon is an Anthropic investor and cloud provider while also competing in AI models.
“A “highly credible trusted partner” found a jailbreak of Fable’s guardrails.”
— David Sacks, White House AI adviser, via X
Cybersecurity for Business: Organization-Wide Strategies to Ensure Cyber Risk Is Not Just an IT Issue
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evidence Still Behind Closed Doors
It is not yet clear what the jailbreak actually does, how reliably it can be reproduced, whether it is unique to Fable, or whether it restores the level of cyber capability Sacks described. The identity and methods of the trusted partner have not been officially disclosed.
Amazon’s reported role also remains unsettled. The source material cites reporting that Amazon flagged the risk and that CEO Andy Jassy had contact with the administration, but Amazon has not confirmed the specific account. That leaves open whether the intervention was driven by a technical safety finding, policy pressure, commercial rivalry, or some mix of those factors.
RC Digital Servo Tester 6 Channels Motor Servo Controller Centering Tool with Over-Current Protection & 2 Control Modes for RC Car Airplane Robots Tester Tool
Specifications: 76mm*53mm(2.99in*2.09in); Weight: 37g (1.31oz)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Patch Timing May Shape Views
The next signal will be whether the restriction is lifted after a quiet patch, remains in place during a longer standoff, or is followed by a more public technical process. A rapid fix could support the view that the issue was concrete and addressable, while a prolonged dispute could strengthen Anthropic’s argument that the government has not shown enough evidence for the action.
Independent review would give readers, developers and policymakers a firmer basis for judgment. Until then, the public record consists mainly of claims by Sacks, Anthropic’s rebuttal and reporting about a possible Amazon role.
AI jailbreak detection tools
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?
The U.S. government restricted Anthropic’s most powerful Fable models after officials said a jailbreak could bypass guardrails and restore dangerous cyber capability. Anthropic disputes the severity of the issue.
What does David Sacks claim?
Sacks says a trusted partner found a serious jailbreak, the administration asked Anthropic to fix or withdraw the model, and the restriction followed after the company refused.
What does Anthropic say in response?
Anthropic says the government did not provide specific technical details and that the reported flaw is minor, already known and reproducible on other public models.
Was Amazon involved?
The source material cites reporting by Semafor, carried by other outlets, saying Amazon may have flagged the issue. Amazon has not confirmed the specifics described in that reporting.
What is still unknown?
The public has not seen the technical evidence, test method, partner identity, or independent review needed to verify either side’s account.
Source: Thorsten Meyer AI
