Anthropic Launches Claude Fable 5, Then Suspends It Days Later on a US Government Order

If you only follow AI news loosely, you may have missed one of the more dramatic weeks the industry has had in a while. In early June 2026, Anthropic, the company behind Claude, launched what it called its most capable publicly available model, Claude Fable 5. Then, just three days later, it had to switch that same model off because of a US government order. I want to walk through what actually happened, in plain language, and explain why it matters even if you are running a small business here in Sabah.

What actually happened

On 9 June 2026, Anthropic released Claude Fable 5, a public version of its more powerful internal model line known as Mythos. Anthropic described Fable 5 as its most capable widely released model, with strong results across software engineering, knowledge work, vision and scientific research. To make a model that powerful safe for general use, Anthropic built in safeguards that block responses in high-risk areas such as cybersecurity and biology, and route a small number of sensitive requests to its older Claude Opus 4.8 model instead.

Then, on 12 June 2026, things changed quickly. The US government, citing national security and export-control authorities, issued a directive ordering Anthropic to suspend all access to Fable 5 and the underlying Mythos 5 model for any foreign national, whether inside or outside the United States. According to reporting from TechCrunch and CNBC, officials acted after learning of a technique that could bypass Fable 5's safeguards and reach the more powerful cybersecurity abilities of the Mythos model underneath. Anthropic has said it believes that workaround was narrow rather than universal, and that similar techniques exist for other widely available models too.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance., Anthropic, official statement, 12 June 2026

Because the directive covered foreign nationals everywhere, including Anthropic's own non-citizen employees, the company said it had little practical choice but to turn the two models off for everyone. So a model that launched as the most capable Claude available to the public was pulled within days of going live.

The part that matters most for you

Here is the reassuring bit, and it is important. This suspension applied only to the two top-tier models, Fable 5 and Mythos 5. Anthropic was clear that its other Claude models were not affected. That includes the Claude Opus, Sonnet and Haiku models, such as Claude Opus 4.8, which most people and businesses actually use day to day. In other words, if you log in to Claude to draft an email, summarise a document, or help write a proposal, your experience did not break. The everyday Claude that we teach in our workshops kept running normally.

So this was not a case of Claude going down or becoming unusable. It was the very newest, most powerful experimental tier being switched off for legal and safety reasons, while the practical, business-ready versions carried on.

Why this matters for Sabah and Malaysian businesses

You might fairly ask: why should a cafe owner in Kota Kinabalu or a trading office in Penampang care about a top-tier AI model being switched on and off in the United States? A few reasons.

• AI tools can change fast. A model can launch, get updated, or be pulled with very little warning. If your whole business workflow depends on one specific cutting-edge feature, you are exposed when it shifts. It pays to stay a little flexible.
• The newest is not always the one you need. The headline-grabbing 'most capable' model is rarely the one a normal business should build on. The stable, widely available versions are usually more than enough for writing, customer replies, summaries and admin, and they are far less likely to be disrupted.
• Safety and regulation are now part of the story. This episode is a reminder that powerful AI is increasingly governed by safety rules and government oversight. That is generally a good thing, but it means the AI landscape will keep moving, so understanding the fundamentals matters more than chasing any single tool.

The practical takeaway

My honest advice has not changed because of this news. Do not get distracted by the race for the biggest, newest model. For almost every Sabah business, the goal is simple: learn to use a reliable AI assistant well, build a few solid habits around it, and apply it to real tasks that save you time every week. The stable Claude models, along with tools like ChatGPT and Gemini, are already powerful enough to make a genuine difference to how you work. What separates businesses that benefit from AI from those that do not is rarely access to the very latest model. It is knowing how to actually use the tools you already have.

If you want to build that practical confidence rather than just read the headlines, that is exactly what we do at NextGen Training Academy. We run hands-on, beginner-friendly, HRDC-claimable AI workshops in Penampang, Sabah, using the everyday version of Claude on real business tasks. You bring your actual work, and you leave knowing how to put AI to use, no matter which model happens to be making the news that week.