This website uses cookies

Read our Privacy policy and Terms of use for more information.

AI Governance
From launch to blackout in days: a government just switched off Fable 5
Anthropic's most capable public model went dark worldwide this week, by an order the public will never see. Here is what happened, and why it matters for anyone building on frontier AI.
 
On 12 June, the US government issued an export control directive restricting foreign nationals' access to Anthropic's two most capable models, Fable 5 and Mythos 5. Because Anthropic cannot cleanly separate who uses what, the practical effect was total. Both models went dark for every customer worldwide, including US firms and Anthropic's own foreign-national staff. Every other Claude model stayed live.
The stated trigger was a single alleged jailbreak. In plain terms, someone asked the model to read a codebase and fix software flaws. Anthropic's reply is worth reading closely. It says that capability is already available in other deployed models, including OpenAI's GPT-5.5, that no universal jailbreak has been found, and that perfect jailbreak resistance is probably not achievable by anyone today. Its safeguards, it argues, were the strongest of any model yet shipped.
Two views, both serious
One view. The government moved fast on a national security signal it could not fully disclose, and caution on frontier capability is exactly its job. Pull access, then verify.
The other view. A narrow, replicable finding was used to recall a model serving hundreds of millions, with no published technical basis. Apply that bar across the industry and you could freeze every frontier deployment, including the defensive security tools that keep the rest of us safe.
Both cannot be right. But notice what neither side disputes. Access to critical AI can be withdrawn overnight, worldwide, with no reasoning you ever get to see. If your operating model assumes the AI you depend on today will be there tomorrow, that assumption just failed a live test. Portability, multi-provider design and governance stopped being architecture niceties. They are continuity controls now.
If your AI provider went dark tomorrow, how many days until your business recovered?
Source
Anthropic's full statement on the US government directive.
Read the statement
 
By John McGann, Zymbos Intelligence
Zymbos Intelligence
zymbos.ai
You are receiving this because you subscribed at zymbos.ai
© 2026 Zymbos Intelligence · John McGann · London, UK
Zymbos Ltd · Company No. 16198848 · Teddington, England