In 2026, nearly every piece of software you touch is racing to bolt on an AI assistant. Meanwhile, the one office suite that announced it had no AI at all just broke its own download records, passing 1 million in a single week and climbing past 2 million.
That’s LibreOffice, the free, open-source alternative to Microsoft Office. Version 26.8 shipped on August 26, 2026, and The Document Foundation, the nonprofit behind it, put the absence of generative AI right in the announcement. Their line was blunt and a little cheeky: yes, no AI is now a feature.
If you’ve been following along at agent101.net, you know I spend most of my time explaining what AI agents do and why they’re useful. So let me be clear about why this story caught my attention. It isn’t an anti-AI victory lap. It’s the clearest signal yet that regular users have started asking a question most software companies skipped over.
What “no AI” actually means here
The announcement came with three specific claims about LibreOffice 26.8, and they’re worth separating because they mean different things:
- No generative AI features. There’s no chatbot in the sidebar offering to rewrite your paragraph or summarize your spreadsheet.
- Your documents don’t go to remote servers. What you type stays on your machine.
- No network access required to function. Unplug the internet and the software still works completely.
That third point is the quiet one, and it’s the one I’d underline for anyone who isn’t technical. Most AI features in consumer software don’t run on your laptop. They run on someone else’s computers. For the assistant to help you, your text has to travel somewhere, get processed, and come back. That’s not a scandal, it’s just how large models usually work. But it changes what a word processor is. A tool that needed nothing but your hard drive becomes a tool that needs a connection, an account, and a company on the other end.
Not a rejection, a set of conditions
The Document Foundation was careful about this part, and the nuance matters. They didn’t say AI is bad. They said they don’t completely reject it. Their position is that office software used by tens of millions of people has to meet certain principles before AI gets added.
Think about what that responsibility looks like. LibreOffice sits on the machines of students, nonprofits, government offices, small businesses, and people in places where bandwidth is expensive or unreliable. When your user base is that broad and that varied, shipping a feature that quietly routes documents to a third party isn’t a product decision. It’s a decision about everyone’s data at once.
That’s a different calculation than a startup makes. A startup can ship, learn, and patch. A foundation maintaining infrastructure for tens of millions of users is closer to a utility. Utilities move slowly on purpose.
It’s also not the only way to handle this. Collabora, which builds a commercial version of the same underlying code, added AI as an opt-in feature that’s switched off by default in Collabora Office 26.04. Same codebase family, two different answers. Both are defensible. One treats AI as something you turn on deliberately; the other treats it as something that needs to clear a higher bar first.
Why the download numbers are the real story
Over 2 million downloads says something that no survey captures as well. People didn’t just tolerate the absence of AI. They actively sought it out. The announcement functioned as marketing, and it worked better than most feature launches do.
My read is that this reflects fatigue rather than hostility. Users have been handed AI in places they never asked for it, often with unclear settings and no obvious off switch. When a product says plainly what it does and doesn’t do, that clarity feels like relief. The appeal wasn’t “AI is useless.” It was “finally, someone told me straight.”
The useful takeaway for the rest of us
You don’t have to pick a side here. You do get a better set of questions to ask about any tool that offers you an AI assistant:
- Does my content leave my device, and where does it go?
- Does the software still work if I’m offline or decline the feature?
- Can I turn it off, and does off actually mean off?
- Who is accountable if something goes wrong with my data?
Those are the same questions The Document Foundation is asking on behalf of its users. The difference is they have the use of a nonprofit with no quarterly growth target, so they can wait for good answers.
For a lot of work, an AI assistant genuinely saves time, and I’ll keep writing about how to use agents well. But a tool that respects a user’s choice to opt out isn’t behind the curve. The 2 million people who went looking for exactly that have made the point better than any argument could.
🕒 Published: