TL;DR
White House AI adviser David Sacks says Anthropic’s Fable model was restricted because a trusted partner found a jailbreak that could restore Mythos-level cyber capability. Anthropic says the government gave little technical detail and that the issue was narrow and common across public models. The evidence needed to settle the dispute remains non-public.
White House AI adviser David Sacks has publicly defended Washington’s export-control action that forced Anthropic’s Fable and Mythos models offline, saying a trusted partner found a jailbreak that could restore cyberweapon-level capability; Anthropic disputes that account, saying officials provided no specific technical detail and that the flaw was narrow and common across public models.
Sacks, co-chair of the President’s Council of Advisors on Science and Technology, said in a June 13 statement on X that a “highly credible trusted partner” found a way around Fable’s guardrails. He said the administration asked Anthropic CEO Dario Amodei to fix the issue or pull the model, and that export controls were issued “reluctantly” after Anthropic refused.
Anthropic’s published response, dated June 12, gives a sharply different account. The company said the government did not provide specific technical detail, described the demonstration as involving a few minor and already-known flaws, and said similar outputs could be produced by other public models, including GPT-5.5, without a special bypass.
Separate reporting by Semafor and other outlets has said the trusted partner may have been Amazon, with CEO Andy Jassy reportedly in contact with the administration. Amazon has not publicly confirmed the specifics cited in the source material. That possible role matters because Amazon is at once an Anthropic investor, a cloud provider for the company, and a competitor in AI models.
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 Tests AI Oversight
The dispute is a major test of how the U.S. government will police frontier AI systems when officials believe a model can aid advanced cyber operations. If the government can cut off access based on non-public evidence, companies, customers, foreign partners, and security researchers may have limited ways to challenge or verify the decision.
The case also shows how safety arguments can serve several interests at once. The government is citing safety to justify a forceful intervention in commercial AI. Anthropic, which has previously warned that powerful models can present cyber risks, is now arguing that the danger in this case is overstated. A reported Amazon role adds another layer because a safety tip could also affect a rival’s flagship launch.
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 Case Comes Back
The dispute centers on the relationship between Mythos and Fable. Sacks’ account treats Fable as a guarded version of Mythos, meaning a guardrail failure could expose Mythos-class cyber capability to users who should not have it. Anthropic’s account treats the reported weakness as narrower and not unique to its systems.
The fight follows a broader policy debate over whether the most capable AI models should face export controls, safety tests, or usage restrictions when they can support cyber work. Anthropic has been one of the companies most associated with arguments for stronger frontier-model oversight. The current case puts that same safety framework on Anthropic itself.
“A “highly credible trusted partner” found a jailbreak, and the order was issued “reluctantly.””
— David Sacks, White House AI adviser, in a June 13 statement on X
AI model jailbreak detection software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Evidence Still Behind Closed Doors
The public has not seen the technical evidence behind the government’s decision. There is no public methodology, no CVE, no named independent assessor, and no reviewable test showing whether the alleged jailbreak is severe, narrow, reproducible, or already present in comparable models.
It is also unclear whether Amazon was the partner cited by Sacks. Multiple reports have pointed to Amazon, but the company has not confirmed the specifics in the source material. Without the underlying test, readers are left with competing claims from parties that each have institutional, commercial, or regulatory incentives.
Volam 4G LTE Cellular Dash Cam with Live Streaming & GPS Tracking – Dual Car Security Camera with Remote Live View,Cloud Storage,AI Motion Detection,Sentry Mode, Anti-Theft Protection for Vandalism
[4G LTE CELLULAR CONNECTIVITY & LIVE STREAMING FROM ANYWHERE] Stay connected to your vehicle 24/7 with our advanced…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Patch Fight Moves To Washington
The next signal is whether the restriction is lifted quickly after a private fix or whether the standoff lasts. A fast restoration after a quiet patch would support the view that officials saw a specific, repairable risk. A longer dispute would give more weight to Anthropic’s argument that the alleged flaw was broad, hard to eliminate, or not unique to its models.
The larger question is whether the government creates a transparent review process for frontier-model safety disputes. Independent technical review would give officials a way to act on real risks while giving companies and users a way to test claims that now sit behind closed doors.
AI-Powered Safety: Streamlined EHS Operations for Managers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What happened to Anthropic’s Fable and Mythos models?
The U.S. government imposed export-control restrictions that led Anthropic to pull access to its most powerful models. Officials say the move was tied to cyber-safety concerns; Anthropic says the reported flaw was narrow and overstated.
Who is David Sacks in this dispute?
Sacks is a White House AI adviser and co-chair of the President’s Council of Advisors on Science and Technology. His June 13 statement is the most detailed public government-side account in the source material.
What does Anthropic dispute?
Anthropic disputes the severity of the alleged jailbreak and says officials did not provide specific technical detail. The company also says similar behavior can be reproduced in other public models.
Was Amazon involved?
Semafor and other outlets have reported that Amazon may have flagged the issue, but Amazon has not confirmed the specifics cited in the source material. That possible role is sensitive because Amazon is an Anthropic investor, cloud provider, and AI competitor.
Why does this matter to AI users?
The case may set a precedent for how governments restrict advanced AI models based on security claims. It affects model access, cybersecurity work, competition among AI providers, and public trust in safety decisions made without reviewable evidence.
Source: Thorsten Meyer AI
