You’re Renting, Not Owning: What the Anthropic Shutdown Teaches Writers

This entry is part 9 of 9 in the series AI for Doubters
TL;DR: The U.S. government switched off Anthropic’s most powerful AI models on a Friday night, proving in one stroke that nobody who uses AI actually owns it. If your business runs on a rented tool, the people who control that tool can change the deal or shut it off without asking you. The fix is old and it works: keep your own books, your own platform, and your own voice at the center, and put a swappable layer between yourself and every service you don’t own. See why owning your sales channel matters.

The most powerful AI model in the world got switched off on a Friday night. Not because it broke. Because the government sent a letter.

On June 12, the Commerce Department told Anthropic that its two newest models, Fable 5 and Mythos 5, could no longer be used by any foreign national. Not just people overseas. Anyone who isn’t a U.S. citizen, including foreign-born people working at Anthropic itself. There was no clean way to block only those users, so Anthropic shut the models off for everyone. Every other Claude model kept running. Just the top tier went dark.

The reason, as far as anyone can tell, was a security worry. Another company claimed it found a way to jailbreak Mythos and pull cyberattack tricks out of it. The government got nervous and dropped the hammer. Anthropic pushed back hard. They said the jailbreak was narrow, that it amounted to asking the model to read code and fix flaws, and that other public models do the same thing without anyone yanking them off the shelf. Was the government right to swing this big? I don’t know. It has the look of a sledgehammer aimed at a thumbtack. But that’s the government for you, and arguing about whether they overreacted skips past the part that should matter to you.

Here’s the part that matters. If your business runs on an AI tool, you don’t own that tool. You rent it.

You’ve Been Renting the Whole Time

When you subscribe to Claude or ChatGPT or any of these things, you’re renting access. Same as Netflix. Same as your phone plan. The company can change the deal whenever it wants, and you get a say in exactly none of it.

I’ve watched this happen plenty of times already, in small ways. I’ve used older Claude models that simply went away. I didn’t ask for them to disappear. One day they were there, the next day I got moved to a newer model whether I liked it or not. In my case the newer ones were better, so I came out ahead. But that was luck, not control. The company decided. I went along with it because I had no other choice. If you want to see how much these tools differ when you’re forced to switch, I broke down the main ones in Claude versus ChatGPT versus Grok.

The Anthropic shutdown is the same thing wearing a bigger coat. A model that hundreds of millions of people used got pulled overnight. This time the order came from the government instead of the company, but the lesson lands in the same spot. The thing you depend on can vanish, and the decision happens somewhere you can’t reach.

I pay two hundred dollars a month for the top Claude plan. I use it hard. Some months I’ve come close to the ceiling. I get real value out of it, and I’m not complaining about the price or the product. But I’m clear-eyed about what I’m buying. I’m buying access that someone else controls. As long as I remember that, I can plan around it. The trouble starts when people forget it.

If you think a software company is different from the government here, ask anyone whose account got frozen overnight. I wrote about what happens when LinkedIn restricts your account with no warning. Different platform, same gut punch. You built something on top of a service, and the service decided you were done.

The News Was Loud Because Loud Sells

Before we go further, a word about how this story got told. A narrow jailbreak, the kind of thing that happens all the time in this field, got turned into a five-alarm fire. The headlines treated it like the end of the world. It wasn’t. It was a normal event in a young industry, blown up to raise your blood pressure.

That’s not a knock on this one story. It’s how every story works now. To earn a click, the writer has to put something in there that spikes your attention. So a routine security flap becomes a crisis, and a corporate disagreement becomes a war. This isn’t new and it isn’t a conspiracy. People have been juicing their stories since before they had paper to write on. The trick is to read the loud version, find the boring fact buried inside it, and act on the fact instead of the volume.

The boring fact here is simple. A powerful tool got turned off for a while. It’ll come back, or it won’t. Either way, smart people don’t build their livelihood on a switch that somebody else can flip.

What This Means If You Write With AI

If you use AI to help you write, this whole episode means almost nothing to your daily work. A more powerful model went on hold. You still have a dozen tools that do the job. You open one up and keep going.

The takeaway is to stay tool agnostic, and at the level most writers work, that’s easy. Prompting Claude and prompting ChatGPT feel about the same. You ask for what you want in plain language, you get a draft, you fix it. Switching from one to the other costs you almost nothing because the skill carries over. Using AI as a real tool instead of a crutch is its own skill, and I covered the honest version of that in how an AI-assisted book actually works.

That stops being true the deeper you go. The minute you build specialized prompts that only work in one tool, or workflows that lean on one company’s quirks, you’ve tied yourself down. Now switching costs real effort. You have to test the new tool, rebuild your prompts, and figure out what broke. Staying tool agnostic is harder than it sounds, because the easy path always runs toward getting locked in. Every shortcut you take inside one platform is a small chain you’ll have to cut later. The work of staying free is real work, and most people skip it until the day they can’t.

The Real Lesson: Build a Wall Between You and the Tool

Here’s where this gets practical, and here’s the rule I’ve lived by for years. Never wire a tool you don’t own straight into the guts of your business. Always put a layer between yourself and the thing you’re using, so you can swap it out without breaking everything you built.

There’s a difference between using AI as a tool and baking AI into your product. Using it as a tool is safe. I prompt Claude to help me draft something, and if Claude vanished tomorrow, I’d prompt something else and lose nothing but a little time. Baking it in is the dangerous version. If your website calls an AI model every time a visitor loads a page, and that model’s company changes its prices or its rules or its whole business overnight, your site breaks and you find out from your customers.

The fix is the same fix I use for everything I don’t control. I keep my own books, my own website, and my own voice at the center, and I treat every outside service as a part I can replace. A lot of entrepreneurs are pouring their content and their authority into AI right now. That’s fine, as long as the foundation underneath is theirs. Use AI to help build your platform. Don’t let AI be the platform. The whole reason a book works as a business asset is that you own it outright, which is also why expertise without authority gets you nowhere and why selling direct beats renting someone else’s storefront.

This principle has a name and a method, and it’s the spine of how I keep my whole operation from breaking when a service disappears. I’ll lay out exactly how I do it, with real examples from my own site, in a follow-up article. For now, the point is the mindset. Assume every tool you rent will someday be taken away. Build so that when it happens, you change one thing in one place and the rest of your work keeps running.

Because it will happen. Maybe not to you, maybe not this year. But the tension between governments and companies and the people who use these tools is just getting started. There are friendly governments and hostile ones, friendly companies and hostile ones, and they’re all going to keep fighting over who’s allowed to touch what. The Anthropic shutdown isn’t the end of that story. It’s close to the beginning. The writers who come out fine are the ones who never confused renting with owning.

Work With a Professional Who Understands the Difference

If you’re ready to build authority on something you actually own, that’s what I do. I help owners, executives, and founders turn decades of hard-won knowledge into books that carry their name and their voice, anchored in a platform no company can switch off.

Visit The Writing King to see more work, or contact me directly to get started.

Get the Free Guides

Join the list and get my condensed books, free. No spam, unsubscribe anytime.

By subscribing you agree to receive occasional emails. Unsubscribe anytime.

People Also Ask

Why did the government shut off Anthropic’s AI models?
On June 12, the Commerce Department issued an export-control directive barring foreign nationals from using Anthropic’s Fable 5 and Mythos 5 models, citing national security. Because there was no way to block only foreign users, Anthropic disabled both models for everyone. The trigger was a reported jailbreak that could pull cybersecurity capabilities out of the model. Anthropic disputed the severity, arguing the technique was narrow and that other public models can do the same thing. Every other Claude model stayed online.
Is it safe to build my business on AI tools?
Using AI as a tool is safe. Baking it into the guts of your product is the risk. When you subscribe to an AI service, you’re renting access that the company, or even a government, can change or cut off without your say. The safe approach is to keep your own books, website, and voice at the center, use AI to help build them, and never let an AI model become the thing your business cannot run without.
What does it mean that AI access is “rented, not owned”?
It means you don’t control the tool. The company sets the price, the rules, and the availability, and they can move you to a different model or shut one down whenever they choose. Older AI models get retired all the time, and users get pushed to new ones whether they want to or not. The Anthropic shutdown showed the same thing at scale, with a government order rather than a company decision behind it.
How do I protect my business from a tool getting shut down?
Put a swappable layer between yourself and any service you don’t own, so you can replace it in one place without breaking everything that depends on it. Stay tool agnostic where you can, since prompting one major AI model feels much like prompting another. Anchor your authority in assets you own outright, like your own books and your own website, rather than in a platform someone else controls.
Should writers worry about the Anthropic model shutdown?
For day-to-day writing work, no. A top-tier model went on hold, but a dozen other tools still do the job, and the skill of prompting carries across them. The real lesson isn’t about this one event. It’s that any tool you rent can be taken away, so your livelihood should never depend on a single switch that someone else can flip.
What’s the difference between using AI as a tool and depending on it?
Using AI as a tool means you prompt it for help and could switch to another tool tomorrow with no real loss. Depending on it means your product or platform calls that specific model to function, so a price change, a rule change, or a shutdown breaks what you built. The first is a convenience. The second is a single point of failure you don’t control.

Related: professional ghostwriting services

When you subscribe to an AI tool, you’re renting access. The company can change the deal whenever it wants, and you get a say in exactly none of it.
Share on X

📝 Disclaimer

The views and opinions expressed in this blog post are solely those of Richard Lowe and are based on personal experience and research. This content is for informational purposes only and should not be construed as professional legal, financial, accounting, or business advice. Always consult with qualified professionals before making important business or legal decisions. Richard Lowe is not a lawyer, accountant, or licensed professional advisor, and this content does not establish any professional relationship.