If you’ve been following AI news, you’ll have heard of Fable 5, one of the most capable AI models released so far. And if you’ve tried to use it recently, you’ll have found it isn’t available. Here’s what’s going on.
What is Fable 5?
Fable 5 is an AI model from Anthropic, the company behind Claude. It was built with especially strong safeguards around sensitive areas, cyber security in particular, and went through extensive testing with government agencies, independent third parties and Anthropic’s own teams before launch. Anthropic says its safety measures were more effective than any model it had deployed before.
Mythos 5 is a related Anthropic model, also now unavailable. It was only offered to a small group of trusted partners before Fable 5 was released. Both sat right at the frontier of what AI can do.
Why can’t you use it?
On 12 June 2026 the US government issued an export control directive ordering Anthropic to suspend access to Fable 5 and Mythos 5 for all foreign nationals, inside or outside the US, including Anthropic’s own foreign national staff. Because there’s no practical way to do that selectively, Anthropic switched access off for everyone.
The government’s stated concern is that a way of getting around Fable 5’s safeguards (a “jailbreak”) has been found. Anthropic received the directive at 5:21pm US Eastern time on 12 June, with no detail about the national security concern.
What does Anthropic say?
Anthropic has been quite direct. When it launched Fable 5 it said publicly that no model can be made completely jailbreak-proof. Every AI system can be bypassed in some way under the right conditions. Its approach was to make jailbreaks narrow or hard to pull off, and to monitor actively so attacks are spotted and shut down quickly.
Anthropic reviewed what it believes the directive is based on and came to a different conclusion: the capability in question is available from other public models, including OpenAI’s GPT-5.5, and is used regularly by security professionals for legitimate defensive work.
Timeline
- 9 June 2026: Anthropic releases Fable 5 publicly, with Mythos 5 available to selected partners.
- 12 June 2026, 5:21pm ET: the US government orders Anthropic to suspend all access to both models. Anthropic is given 90 minutes to comply, and access is switched off for every customer.
- 12 June 2026: Anthropic confirms it has complied but disputes the reasons. Dozens of cyber security researchers, AI entrepreneurs and executives sign an open letter criticising the government’s action.
- 21 June 2026: talks between Anthropic and the Trump administration continue. President Trump says they’re “going fine” and that he no longer sees Anthropic as a national security threat.
- 1 July 2026: restrictions are expected to be lifted within 24 hours, subject to an agreement between Anthropic and the US government and strict security controls.
Will it come back?
Anthropic has complied while disputing the basis for the order. It says it believes this is a misunderstanding and is working to restore access as soon as it can. All its other models, including Claude Opus, Sonnet and Haiku, were unaffected.
What it means for your business
For most New Zealand businesses, very little day to day. Fable 5 and Mythos 5 were frontier, research-grade models, not the everyday tools most businesses use. The Claude models doing most of the real work are still fully available.
What it does show is how important it is to build your AI approach on solid ground, so it doesn’t depend on one model or one provider and can adapt when things change. That’s worth thinking about however this particular story ends. If you’d like to talk it through, our sister company Kambium can help, or start with where to start with AI.
Want an AI approach that does not hinge on a single model or provider? Book a free health check, or call us on 0800 688 768.