On June 12, the US government issued an export control directive ordering Anthropic to immediately suspend access to its two most capable AI models – Fable 5 and Mythos 5 – by any foreign national, including Anthropic’s own non-US employees. Anthropic complied but suspended access globally to ensure compliance. By Sunday, June 15, a group of 54 CISOs, cybersecurity practitioners, and vendors had signed an open letter addressed to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross demanding the directive be lifted. YourNewsClub treats the 54-signature open letter as a more significant signal than the export order itself – it represents the working community most directly affected by the ban speaking publicly against a government action that took the tools they use to defend systems away with no prior consultation and no public explanation.
The signatories’ central technical argument is precise. They acknowledge that Fable 5 and Mythos 5 are “quite good at finding flaws and weaponizing exploits.” But they argue that the same capabilities can be replicated on OpenAI’s GPT-5.5, on Anthropic’s own publicly available Claude Opus 4.8 and Sonnet, and on Chinese models including Kimi 2.7. Katie Moussouris, a well-known vulnerability researcher and signatory, stated: “The bugs used to demonstrate the techniques in the paper can be found using the other models. The method in the paper is a guardrail bypass technique. Other models that lack the Fable guardrails often won’t refuse the straightforward request to look for security bugs, so they don’t need a bypass.” That is a specific technical rebuttal: the ban restricts a model with safety guardrails while leaving unrestricted access to models without those guardrails.
The government’s stated rationale is national security concerns tied to cybersecurity vulnerabilities. Anthropic believes the directive originated from Amazon CEO Andy Jassy reportedly warning Trump administration officials that Fable 5’s guardrails were flawed – that a jailbreak technique could exploit the model’s capabilities in ways that posed national security risk. YourNewsClub identifies that account, if accurate, as the most unusual element of the episode: a major customer of Anthropic’s competitor apparently provided the intelligence input that produced a government order restricting Anthropic’s products.
Jessica Larn, who studies macro-level technology policy and infrastructure impact of AI, draws the policy architecture problem plainly: “An export control order on frontier AI models, issued without a public regulatory process, sets a precedent that any AI company can lose access to its customers – globally – based on a classified or informal intelligence assessment. That is not a stable governance framework for an industry where capability advances happen in weeks.” YourNewsClub considers Larn’s framing the most structurally significant concern in the open letter’s argument, because the procedural objection matters regardless of whether the specific technical judgment about Fable 5 was correct.
The open letter calls for “a democratic rule-making process” based on scientific research conducted by industry and academic experts, used “only to the minimal extent necessary to ensure the safety of the American public.” China’s advanced AI models, the signatories note, are “only months behind the best American models.” The uncomfortable implication: if the ban on Fable 5 and Mythos 5 is maintained while adversaries develop equivalent capabilities, the US effectively imposes an asymmetric restriction – limiting its defenders without limiting the threat. Maya Renn, whose work focuses on the ethics of computation and access to power through technology, adds the distributional dimension: “When governments restrict access to powerful AI tools in the name of security, the question of who loses access – defenders, researchers, educators, or only foreign nationals – determines whether the restriction reduces risk or simply concentrates capability in fewer hands.” Three things to watch next: whether Lutnick or Cairncross respond to the open letter, whether Anthropic pursues a legal challenge to the directive, and whether Congress requests a public explanation of the technical reasoning behind the ban. Anthropic stated that the standard the government applied, if maintained, would halt all new frontier model deployments across the AI industry. Your News Club expects this case to become the first major test of whether US export control law can apply to AI model access without a formal rulemaking process.