TL;DR
White House AI adviser David Sacks said Anthropic’s Fable model was restricted after the company refused to address a jailbreak that could restore Mythos-level cyber capability. Anthropic disputed that account, saying the government provided no specific technical detail and that the alleged flaw was minor. The central evidence remains non-public, leaving the severity and process unresolved.
White House AI adviser David Sacks said over the weekend that Anthropic’s most powerful Fable models were restricted after the company refused to fix a jailbreak that could restore cyberweapon-like capability, while Anthropic says the alleged flaw was minor and not unique to its systems.
Sacks, co-chair of the President’s Council of Advisors on Science and Technology, said a “highly credible trusted partner” found a way around Fable’s guardrails. According to Sacks, the administration asked Anthropic chief executive Dario Amodei to fix the issue or pull the model, and Amodei refused. Sacks said the export control was issued “reluctantly.”
Anthropic has rejected that account. In a June 12 blog post described in the source material, the company said the government did not provide specific technical detail, that the demonstration showed only a few minor and already known flaws, and that similar behavior could be reproduced in other public models, including GPT-5.5, without a special bypass.
The two accounts differ on the core technical question. Sacks frames the issue as restoring the operability of a cyberweapon. Anthropic describes it as a narrow potential jailbreak that does not justify recalling or blocking a model used at large scale. No public technical record currently resolves that dispute.
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.
Safety Claims Meet Market Power
The dispute matters because it places AI safety claims at the center of a major government intervention in commercial AI. If Sacks’s account is accurate, the action signals that U.S. officials are willing to restrict advanced models when they believe guardrails fail in ways tied to cyber capability.
If Anthropic’s account is closer to the facts, the episode raises a different concern: that a model can be limited through a process the public cannot inspect, based on evidence that remains secret and technical claims that outside experts cannot test.
The case also carries business consequences. Reporting cited in the source material says Amazon may have been the partner that flagged the issue, though Amazon has not confirmed the specifics. Amazon is described as an Anthropic investor, compute provider and AI competitor, a mix of roles that makes the source of the allegation relevant even if the underlying safety concern is real.
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.
Fable Built From Mythos
The dispute centers on Fable, which Sacks described as Mythos with guardrails. In his account, if those safeguards can be bypassed, Fable would give users access to the same class of cyber capability that made Mythos a concern.
The source material says Anthropic had previously argued that Mythos posed serious cyber risks and should face regulation. That history now cuts both ways: it supports the government’s concern about capability, while also making Anthropic’s objection to the current restriction more visible.
The episode follows a broader fight over who sets the safety standard for frontier AI systems: companies developing the models, government officials responding to security warnings, or independent reviewers able to test the claims.
Hacking and Security: The Comprehensive Guide to Ethical Hacking, Penetration Testing, and Cybersecurity (Rheinwerk Computing)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evidence Still Behind Closed Doors
It is not clear what prompt, tool chain, evaluation method or exploit path was used in the reported jailbreak. The government has not published a technical methodology, a named independent review or a public vulnerability record.
The identity of the trusted partner also remains unconfirmed. The source material cites reporting by Semafor, carried by Fortune and others, saying the partner may have been Amazon and that Amazon chief executive Andy Jassy was reportedly in contact with the administration. Amazon has not confirmed the specifics cited in the source material.
It is also unresolved whether the alleged flaw is specific to Fable, present across many public models, or serious enough to justify the government action. At this stage, both Sacks’s account and Anthropic’s response are claims from interested parties.
AI model jailbreak detection tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Patch Timing Becomes Test
The next milestone is whether the restriction is lifted quickly after a private fix, or whether the standoff continues. A fast reversal after a quiet patch would support the view that the government found a concrete defect Anthropic could address. A longer dispute would strengthen questions about whether the flaw was severe, specific and reviewable.
Outside researchers and policymakers are likely to focus on process: whether future AI safety actions rely on public technical standards, independent review and clear disclosure rules, or continue through private warnings and competing public statements.
The Confidence Advantage: Optimizing Privacy, Cybersecurity and AI Governance for Growth
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened between David Sacks and Anthropic?
Sacks said the U.S. government restricted Anthropic’s Fable models after a trusted partner found a jailbreak and Anthropic refused to fix or pull the model. Anthropic disputes the severity and says the government did not provide specific technical detail.
What is Fable?
Fable is described in the source material as Anthropic’s powerful model family tied to Mythos-level capability with guardrails. Sacks argues that bypassing those guardrails could restore dangerous cyber capability.
Did Amazon report the alleged jailbreak?
The source material cites outside reporting that Amazon may have been the trusted partner, but says Amazon has not confirmed the specifics. That remains unverified in the material provided.
Is the jailbreak confirmed?
A jailbreak allegation is confirmed as a claim made by Sacks, and Anthropic has confirmed it disputes the government’s framing. The technical details needed to verify severity have not been made public.
Why does this matter beyond Anthropic?
The dispute may shape how governments restrict frontier AI systems, how companies contest safety claims, and whether future model interventions can be checked by independent technical review.
Source: Thorsten Meyer AI
