LibreOffice Just Declared “No AI” a Feature — and Smashed Every Download Record in Its History. Here’s What Your Subscription Office Suite Is Really Selling.
In a software world where every vendor is currently racing to stuff a chatbot into your document editor, one nonprofit did the exact opposite — and the public reacted the way they only ever react when someone finally stops annoying them: they flooded the download page. LibreOffice just declared “no AI” a feature, and it may have just sold more copies of a single release than ever before. That is not a small thing. It is certainly not an accident. And the story underneath the headline is far more interesting than “spreadsheets, but no robot.”
The Headline That Wasn’t Actually New
Let’s get the timeline straight, because getting it wrong is exactly what the hype machine wants. LibreOffice 26.8 shipped on August 26, 2026, and its announcement already carried a section literally headed “No artificial intelligence,” stating the release “contains no generative AI features” and calling it “a deliberate design position.” Then on September 3, The Document Foundation (TDF) — the nonprofit that owns the project — published a blog post titled “Yes, no AI is now a feature,” turning a release note into a full statement of principles. It wasn’t until October 6 that TechCrunch’s security editor Zack Whittaker dragged it in front of a much wider audience.
So before you skim this thinking it broke this morning, know that the policy is roughly a month old. That actually makes it more impressive, not less. Most AI controversies live for a news cycle and die. This one was stated, defended, and then quietly compounded by the single strongest signal available in software: people literally opening their wallets — or rather, refusing to pay anything at all — and downloading by the millions. The story isn’t the announcement. It’s what happened after nobody asked for it.
What They Actually Said (And What They Deliberately Didn’t Say)
Here is where most coverage gets it subtly wrong. LibreOffice is not banning AI out of luddite spite, and it is not declaring that generative models are worthless. It is building a gate with six published conditions, and it is being candid that no integration on the market today clears all six. The foundation was careful to spell out what this is NOT: the criteria “do not amount to a definitive rejection,” and the current verdict is explicitly “an assessment of the current state of the technology,” not “a judgement on the value of the sector.” That restraint is a deliberate choice by a group that knows it could easily have written a takedown of the whole industry and gotten away with it.
The heart of the September post is a checklist that reads less like a manifesto and more like an IT procurement rubric. Before any AI feature can be switched on by default, it must satisfy all of the following:
- User-controlled execution. The user decides where inference runs — on their own computer, on infrastructure they directly control, or on a service they independently chose. The foundation is blunt that “a default setting that silently redirects to a single provider is not a choice.”
- No content leaves without authorization. Uploading a document is the user’s decision, made consciously. The application does not quietly exfiltrate your file in the background.
- No telemetry. LibreOffice already collects zero usage data, and this extends that rule straight into the AI layer. No prompt logging, no usage pings.
- No single provider. The integration must sit behind an open interface that any backend can implement. “An integration that only works with a single company’s API is a lock-in mechanism, regardless of how it is described in the release notes.”
- No compromise on format. Generated text is saved inside the Open Document Format (ODF), keeping structure and styles intact, so an assistant can’t quietly smuggle you back into a proprietary file.
- Entirely optional. The feature must be installable, removable, and hidden from the interface for administrators whose policy forbids it — even across thousands of workstations.
Measured against that bar, the foundation says plainly that “there is no integration that meets all the requirements.” And that is the quiet gut-punch for the big vendors: the commercial assistants that power most office suites fail on the first and fourth conditions alone, because they were architected around one company’s cloud from day one. You cannot easily satisfy “no single provider” when your entire product is a single provider.
The line that will live in marketing decks is the most honest sentence in the whole post: “Any organisation handling confidential, legally privileged or personal data needs to know where that data goes, and the only assurance that survives an audit is that it does not leave the machine.” Not “we try our best.” Not “our lawyers say it’s fine.” The only assurance that survives an audit. Say that to a SaaS salesperson about their AI assistant and watch them change the subject.
The Number That Silences the Skeptics
Opinions are cheap. Download counts are not. If you think this “no AI” stance is just a principled posture aimed at a niche of open-source purists, the numbers will correct you fast.
According to Cybernews, which cited the project’s own figures, LibreOffice 26.8 passed one million downloads in its first week — an exact count of 1,031,162. The project called it “the highest number of downloads in the first week of any LibreOffice release.” By early September, TDF founding member Italo Vignoli told Cybernews it marked “the first time in LibreOffice history” with two consecutive weeks above one million downloads. The momentum did not stall: the foundation later pointed out that 26.8 had pulled in more than 4 million downloads within the month of September alone.
Zoom out and the trend is unmistakable. By Vignoli’s numbers, LibreOffice logged 29,755,011 downloads from the start of 2026 through early September — already roughly 20 percent ahead of the entire 2025 total of 24,830,220, which itself edged out 2024’s 23,732,502. Weekly figures climbed from about 718,211 in week 35 to roughly 966,458 in the week ending September 6, a jump of close to 35 percent.
Now, a healthy dose of skepticism is warranted, and the foundation itself offers it. That one-million-in-a-week figure counts only the official download page, which excludes the vast majority of Linux users who grab LibreOffice through their distribution’s package manager — so the real reach is even bigger. And correlation is not causation: Vignoli himself partly credited the spike to back-to-back release announcements and the broader European push to cut dependence on American suppliers. So no, we cannot prove the “no AI” banner is the sole reason. But when the best-selling release in a decade-old project’s history is the one that most loudly refused to add the single hottest feature of the decade, it would be silly not to listen to what the crowd is telling us.
The Real Reason Nobody Wants to Say Out Loud
Here is the uncomfortable truth that makes this story actually about you instead of about a spreadsheet program: LibreOffice’s stance has almost nothing to do with the technology and almost everything to do with the business model. The foundation was remarkably candid about this. An AI assistant, it explained, is a perfect justification for a subscription company: it “justifies a price increase and strengthens the case for keeping all documents within its own infrastructure.” Then came the line that lands like a slap: the Document Foundation, by contrast, “has no subscription tiers to protect, no upsells, no data to monetise,” which is what lets it take its time deciding whether something is genuinely useful “as we are not driven by quarterly results.”
Read that again, because it describes two completely different companies pretending to sell the same product. One of them adds a chatbot, raises your bill, and quietly gains the ability to see everything you type. The other adds better typography and charges you nothing. The price history says it all. In early 2025, the big suites rolled out AI and raised prices across the board:
- Microsoft 365 Personal: $6.99 to $9.99 per month — a $3 jump, or +43 percent — with Copilot and 60 AI credits a month bolted on.
- Microsoft 365 Family: $9.99 to $12.99 — another +30 percent.
- Google Workspace Business Starter: $6 to $7 per user per month — +17 percent.
- Google Workspace Business Standard: $12 to $14 — +17 percent.
- Google Workspace Business Plus: $18 to $22 — +22 percent.
- LibreOffice: $0 to $0. No change. No AI. No problem.
The percentage math is what stings. A $3 increase on a $6.99 plan is not “a small bump”; it is a 43 percent hike, and it was justified to you by the word “AI.” Meanwhile your free, open-source alternative improved its paragraph settings and stayed free. That is not a coincidence. That is a business model telling you exactly what it thinks your documents are worth. When a subscription company tells you it must add AI to justify its price, it is quietly admitting that AI is the thing it is actually selling you — and that the price was never really about the software.
You’re Already Tired of This (You Just Haven’t Named It)
LibreOffice did not invent this backlash; it just refused to prettify it. For years, the software industry has been slowly packing its products with what critics have started calling “enshittification” — the slow decay of a good tool into a bloated shell of sticky chatbots, pop-up assistants, and “smart” suggestions that get worse for the user while apparently getting better for the shareholders. The public has absorbed this for a long time. They are now pushing back, and LibreOffice is betting that a genuinely large slice of that frustration is on the side of “just give me the document editor.”
The rest of the industry is hearing the same rumbling. Mozilla moved to let users block all of its generative AI features outright. A command-line tool appeared that can quickly strip Apple Intelligence out of macOS entirely. Even as Firefox pushes the other way with a beta “Smart Window” powered by Mistral models, the dominant street sense is that users are fatigued by assistants that were never asked for. That is the tension LibreOffice is exploiting with elegant simplicity: while every competitor is adding more, LibreOffice is adding nothing by default, and in a market saturated with unwanted features, restraint has become the feature.
Think about it like a restaurant. When every diner at the next table is forced to watch a waiter aggressively upsell a dessert they did not order, the person who quietly brings you water and leaves you alone suddenly looks like a saint. LibreOffice is the saint. It is the software that looks sane precisely because everyone else went feral.
When Countries Start Showing Up
Here is the part that turns a software policy into a geopolitical one. LibreOffice was already being adopted by governments before the “no AI” post was ever written, and the two trends stack on top of each other like a compound interest problem. In June 2025, Denmark’s Ministry of Digital Affairs announced it would move its staff out of Microsoft Office and into LibreOffice, following the German state of Schleswig-Holstein, which is migrating roughly 30,000 government workstations to open-source software. The foundation points out that Germany, NATO, and the European Commission all mandate the ODF format for public documents.
Why does this matter for an article about AI? Because digital sovereignty and the “no AI” stance are the same coin. If a hospital, a law firm, or a ministry is worried about American cloud vendors holding their data, an AI assistant that silently routes your medical records or case files to a foreign provider is not a convenience — it is a sovereign risk. LibreOffice’s own framing puts it starkly: its software is “used by tens of millions of people – in schools, hospitals, public bodies, law firms and thousands of other organisations.” For those entities, “the only assurance that survives an audit is that it does not leave the machine.” An AI assistant, by definition, makes your data leave the machine. So the “no AI” policy is really a sovereignty policy wearing a privacy costume, and nations are the ones most likely to notice.
It’s Not A Skinny Release (They Built Stuff Instead)
The easiest insult to throw at a “we added nothing” product is that it is thin. That would be wrong here, and it is worth defending the actual work. LibreOffice 26.8 was not a hollow shell dressed up in a manifesto; it shipped a genuinely useful set of improvements that quietly matter to anyone who formats documents for a living.
Writer gained a Paragraph Composer that balances word spacing across a whole paragraph instead of optimizing each line in isolation — killing that annoying zig-zag look of justified text. OpenType variable fonts are now supported natively, and baseline grids work inside frames. The single largest body of work targeted right-to-left and complex writing systems: Writer now detects paragraph direction automatically, Math gained named functions for the N’Ko and Adlam scripts used for Manding and Fulani languages, and new documents default to “start” alignment rather than “left.” The release also preserves newer Office chart types like waterfall and treemap across a round trip through Microsoft files, even though it cannot yet draw them itself. The follow-up 26.8.1, released October 2, added around 40 fixes, mostly for DOCX, XLSX, and PPTX compatibility, and Calc gained Calculated Fields in pivot tables.
And notice what they spent their engineering energy on instead of a chatbot: typography, writing systems, and file compatibility. Those are the things that make a document editor quietly better every single day. The foundation also notes, matter-of-factly, that “none of this work has a commercial market” — which is precisely why a for-profit company would never build it, and why a nonprofit exists to. LibreOffice runs under the Mozilla Public License v2, carries “no advertising, no telemetry, no subscription, no account requirement,” and is funded by donations and ecosystem companies. In September, the Dutch broker Zerodha’s FLOSS fund handed the project $100,000. The first release even shows a donation banner in the Start Centre instead of a paywall.
What This Means For Your Desk (And The AI You’ll Actually Get)
Before you go delete your Microsoft 365 subscription in a fit of principle, let’s get practical, because the honest answer is nuanced. LibreOffice’s stance does not forbid AI; it just moves the decision from the vendor back to you. The foundation openly points users who want AI toward third-party extensions, most of which connect to a locally running model through Ollama, LM Studio, or any OpenAI-compatible endpoint — meaning your document never leaves your computer and satisfies the very first condition.
But here is the catch that administrators keep missing: an extension pointed at a public cloud API quietly undoes the entire privacy benefit. Install a “LibreOffice + ChatGPT” extension that phones home, and you have just recreated the exact thing the foundation refused to ship by default. So if you want AI in LibreOffice, the responsible path is a local model you control, vetted like any third-party code, with the endpoint confirmed to stay on your machine.
For most people, the realistic middle ground is a mixed estate: LibreOffice for anything sensitive — contracts, medical forms, financial records, government paperwork — and the subscription suite for what it is still genuinely better at, like deep integration into email and calendars that no local extension can match. Neither is a total surrender. And if you are setting this up, note that the older 26.2 branch reaches end of life on November 30, 2026, so any new deployment should standardize on 26.8 from the start.
The bigger takeaway is about where your loyalty should live. The specific assistant in your current suite will be obsolete, sketchy, or shut down within a few years — that is guaranteed. What will not change is the fundamental question of who owns the data you type into a box every single day. LibreOffice’s answer is boring and unglamorous and, frankly, the only one that survives an audit: your documents stay where you put them, and nobody gets to sell the record of what they contained.
So here is the question I want to leave you with: the next time your office suite quietly adds a “helpful” new feature and tacks a few dollars onto your bill, are you going to keep assuming it is doing you a favor, or are you finally going to ask the only question that has ever actually mattered — who is really profiting from your data?