Tonight, across America, the sky's full of fireworks. Independence Day. Meanwhile, in offices and spare bedrooms everywhere else, a quieter thing is playing out. Thousands of people are spending their weekend hammering away at an AI model called Claude Fable 5, trying to get the last bit of use out of it before Monday.
If you missed it, here's the short version. Fable 5 is Anthropic's most capable model, and it's had a month of it. On 12 June, access got pulled without warning after a US government export-control order decided the model was too capable to leave lying around. It came back on 1 July. Then came the sting in the tail: from 7 July it leaves the standard subscription plans. Want it after that, you pay for usage credits or go through the API at prices that made a fair few developers wince.
Anthropic say it's temporary, a capacity problem rather than a change of heart. They might well be right. But watch what people are actually doing this weekend. They're not waiting to find out. They're queueing up their hardest problems, their biggest refactors, their worst documents, and running them through while the tap's still open.
When people binge a tool like it's the last weekend of a closing-down sale, they're telling you something. They built their working lives on it, and they never once planned for losing it.
Now ask the uncomfortable question about your own firm
I work with professional services firms across Greater Manchester, and here's what I know. Somewhere in your business, right now, there's someone whose entire way of working depends on an AI tool. Maybe it's Copilot drafting their emails. Maybe it's ChatGPT summarising documents. Maybe it's something you've never heard of, doing something you don't know about, with data you'd rather it didn't have.
The Fable episode is a preview of a very ordinary future. Tools change price. Tools change terms. Tools get suspended by governments, bought by competitors, or switched off overnight. None of that is in your control. What is in your control is knowing which tools your firm actually depends on, what work flows through them, and what you'd do on the Monday morning one of them disappears.
Most firms can't answer any of those three. Not because anyone's careless, but because nobody's ever been asked to write it down. The dependency builds up one helpful shortcut at a time, the way ivy takes a wall.
What sensible firms are doing about it
The fix isn't dramatic. It's an inventory: which AI tools are in use, by whom, for what, with which data. Once that exists, the continuity question more or less answers itself, and so do several others your insurer and your regulator have started asking. It's the same document that tells you your training needs and your policy gaps. One piece of work, several problems solved.
So enjoy the fireworks, wherever you're watching from. But when the smoke clears, ask your team one question on Monday: if our favourite AI tool went the way of Fable this week, would we actually know what we'd lost?
If the answer's a shrug, you know where to find me.
Update: since this went up, Anthropic pushed the cutoff back to 19 July for Pro accounts. Max and Teams plans kept access throughout. Good news if you're on those tiers, but it doesn't change the point. The tap gets turned off in the end. The only question is whether you know what runs through it.
Find out what your firm depends on
Our AI Risk Assessment maps every tool, every use and every dependency across your business, in plain English. It starts with a free thirty minute call.
Book a free intro call