Hi Reader,

Well, that aged fast.

On Saturday I sent you my honest review of Claude Fable 5. A few days later, the US government stepped in and Anthropic pulled it offline. Same story for Mythos 5.

👉 Read My Original Fable 5 Breakdown​

Here's what actually went down, because the headlines are messier than "banned":

• It started as a foreign-national order, not a full ban. The Commerce Department told Anthropic to cut off all foreign nationals, including its own non-citizen employees. To comply, Anthropic had to shut Fable 5 and Mythos 5 down for everyone.
• The stated reason was narrow. The government flagged one method to get around Fable's safeguards using code review. Anthropic argues it's a single edge case, that GPT-5.5 can do the same thing, and that no model is perfectly jailbreak-proof.
• Follow the incentives. Amazon's CEO warned the White House that Fable could aid cyberattacks. Amazon is both an Anthropic investor and a competitor, and this is happening right before Anthropic's IPO. Read into that what you want.

Anthropic says it's working to restore access, with no timeline yet. Other Claude models are untouched. You can read Anthropic's official statement for more details.

🤔 So What Should You Do Now?

This is the only question that matters to us. And honestly? Don't panic, and don't change much.

Claude Opus 4.8 and GPT-5.5 are still incredibly powerful models and cover 99% of real knowledge work. Marketing, copywriting, research, strategy, etc.

Fable was overkill for most of that anyway, which is exactly what I told you the other day.

But here's what has me thinking.

This is a slippery slope. How does a government decide a model is "too powerful"? How much of that call is real safety versus competitors lobbying behind the scenes?

If Fable crossed the line, does OpenAI's next model or Google's next model cross it too? Nobody has a clean answer yet.

The bigger lesson: you can't fully rely on frontier models when a single order can pull one overnight.

So my move is simple. Lean on Opus 4.8 or GPT-5.5 for now, and start poking at open-source options like Qwen and DeepSeek so you're never stuck if the rug gets pulled again.

This is a pivotal moment for AI, and I'll be watching how it plays out closely.

Where do you think this goes from here? Hit reply and tell me. I read every response.

Appreciate you reading 🫡
Ryan

P.S. This is the whole argument for how I work. Models come and go, your systems shouldn't. My AI Skills Stack is 25+ skills I use to run my business on whatever model is best. One-time fee, lifetime updates.

You can also share your thoughts on this story over in my AI community. This has 1000+ members using AI to level up their work.