What Is Fable 5? Why Can’t You Access It Anymore?

by | Jun 29, 2026 | AI, IT News & Insights New Zealand | Cybersecurity, AI & Microsoft Updates

If you’ve been following developments in AI, you may have heard about Fable 5 — one of the most capable AI models released to date. If you’ve tried to access it recently, you’ll have found it unavailable. Here’s what you need to know.

What Is Fable 5?

Fable 5 is an AI model developed by Anthropic, the company behind Claude. It was designed with particularly strong safeguards around sensitive tasks, notably cybersecurity, and went through extensive pre-launch testing with government agencies, independent third parties, and Anthropic’s own internal teams. By Anthropic’s account, its safety measures were more effective than those of any previously deployed model at the time of release.

Mythos 5 is a related model from Anthropic, also now unavailable. Mythos was made available only to a select group of trusted partners before Anthropic released Fable 5.  Both were positioned at the frontier of what AI models could do.

Why Is It Unavailable?

On 12 June 2026, the US government issued an export control directive ordering Anthropic to suspend access to both Fable 5 and Mythos 5 for all foreign nationals, whether inside or outside the United States, including foreign national Anthropic employees. Because complying with that directive on a selective basis isn’t technically feasible, Anthropic disabled access for all customers.

The US government’s stated concern is that a method of bypassing Fable 5’s safeguards, commonly called a “jailbreak”, has been identified. Anthropic received the directive at 5:21pm ET on 12 June with no specific details of the national security concern.

What Does Anthropic Say About the Jailbreak?

Anthropic has been direct about its position. When it launched Fable 5, the company publicly stated that perfect jailbreak resistance isn’t achievable for any model.  Every AI system in the industry is vulnerable to some form of bypass under specific conditions. Their strategy was to make jailbreaks either narrow in scope or difficult to produce, combined with active monitoring to detect and shut down attacks quickly.

On the specific finding the government has cited, Anthropic reviewed what it believes is the basis of the directive and reached a different conclusion: the capability demonstrated is available from other publicly accessible models, including OpenAI’s GPT-5.5, and is regularly used by security professionals doing legitimate defensive work.

Timeline

June 9, 2026 — Anthropic releases Fable 5 publicly, with Mythos 5 available to select partners.

June 12, 2026, 5:21pm ET — The US government issues an export control directive ordering Anthropic to suspend all access to Fable 5 and Mythos 5. Anthropic is given just 90 minutes to comply. Access is disabled for all customers.

June 12, 2026 — Anthropic publishes a statement disputing the basis for the directive while confirming compliance. Dozens of cybersecurity researchers, AI entrepreneurs, and corporate executives sign an open letter criticising the government’s actions.

June 21, 2026 — Negotiations between Anthropic and the Trump administration are ongoing. President Trump says talks are “going fine” and that he no longer views Anthropic as a national security threat.

July 1, 2026 – Breaking news!!  Restrictions are expected to be lifted in the next 24 hours, subject to an agreement between Anthropic and the US Government and strict security controls being in place.

Will Access Be Restored?

Anthropic has complied with the directive while disputing its basis. The company has said it believes this is a misunderstanding and is working to restore access as soon as possible.  Access to all other Anthropic models, including Claude Opus, Sonnet, and Haiku was unaffected.

What This Means for Businesses Using AI

For most New Zealand businesses, the immediate practical impact is minimal.  Fable 5 and Mythos 5 were frontier research-grade models rather than the everyday tools most organisations are working with. The Claude models powering day-to-day AI work remain fully accessible.

What this situation does illustrate is the importance of building your AI strategy on a solid foundation — one that isn’t dependent on a single model or provider, and that can adapt when the landscape shifts. That’s something worth thinking about regardless of how this particular situation resolves.

If you’d like to talk through how your business can build a resilient AI approach, get in touch with the Kambium team.