LibreOffice AI policy, as The Document Foundation now states it, fits in one sentence: there will be “no AI of any kind in the default installation” of LibreOffice until an integration meets six conditions that, today, none does. The non-profit behind the free office suite set this out in a blog post titled “Yes, no AI is now a feature”, and TechCrunch put it in front of a much wider audience on 6 October 2026.
That makes LibreOffice the odd one out. Microsoft and Google have spent two years building generative AI into their office software and raising prices to match. The Document Foundation has gone the other way, and its newest release, LibreOffice 26.8, has just set the project’s download record.
Below we explain what the LibreOffice AI policy actually says, the six conditions behind it, why “no AI” is being sold as a feature, what the download figures show, how to use AI with LibreOffice anyway, and what the stance means for UK businesses that handle confidential documents.
Table of contents
- What The Document Foundation Actually Said
- The Six Conditions Behind the LibreOffice AI Policy
- Why “No AI” Became a Selling Point
- The Download Numbers Behind the LibreOffice AI Policy
- What LibreOffice 26.8 Ships Instead of AI
- How to Use AI With LibreOffice Anyway
- Privacy, Audits and the LibreOffice AI Policy
- What It Means for UK Businesses
- What Could Change the LibreOffice AI Policy
- LibreOffice AI Policy: Frequently Asked Questions
- References
What The Document Foundation Actually Said
The LibreOffice AI policy did not arrive in one announcement. It built up over six weeks, from a line in a release note to a full statement of principles.
From a release note to a blog post
On 26 August 2026 The Document Foundation released LibreOffice 26.8, the work of 206 contributors, 155 of them volunteers. Its announcement had a section headed “No artificial intelligence“, which said the release “contains no generative AI features” and called this “a deliberate position”. One comment on that announcement asked: “So, no AI is now a feature?”
The foundation answered on 3 September with “Yes, no AI is now a feature”. TechCrunch’s security editor, Zack Whittaker, covered it on 6 October under the headline “LibreOffice says ‘no AI’ is now a software feature”. TechCrunch described the statement as coming “this week”, but the post it links is dated 3 September, so the policy is a month old rather than new.
The words that matter
The core sentence is worth reading in full. An office suite “used by tens of millions of people – in schools, hospitals, public bodies, law firms and thousands of other organisations – will not add a feature to its default configuration until that feature can be delivered under the conditions defined by the project”.
The foundation is careful about what it is not saying. The criteria “do not amount to a definitive rejection”, and the current verdict is “an assessment of the current state of the technology”, not “a judgement on the value of the sector”. The LibreOffice AI policy is a gate with published conditions, not a ban.
Where the line is drawn
The LibreOffice AI policy covers the default installation only. LibreOffice still runs extensions, and the foundation openly points users who want AI towards them. So the practical rule is this: nothing generative ships in the box, and anything you add is your decision, made deliberately.
The Six Conditions Behind the LibreOffice AI Policy
The heart of the September post is a list of six conditions an AI feature must meet before it can be enabled by default. They read like a procurement checklist, and that is how a business should use them. Every other part of the LibreOffice AI policy follows from this list.
| Condition | What TDF requires | What it rules out |
|---|---|---|
| User-controlled execution | The user picks where inference runs: own computer, own infrastructure or an independently chosen service | A default that silently routes to one provider |
| No content leaves without authorisation | Sending a document is the user’s decision | Background uploads by the application |
| No telemetry | No usage data, for AI features as for everything else | Prompt and usage logging |
| No single provider | Open interfaces any backend can implement | An integration tied to one company’s API |
| No compromise on format | Generated content saved as ODF with structure and styles intact | Assistants that output proprietary formats |
| Entirely optional | Installable, removable and hidden for anyone who does not want it | AI buttons administrators cannot remove |
Control over where inference runs
The first two conditions of the LibreOffice AI policy are about location. The user must be able to choose whether a model runs “on the local computer, on an infrastructure directly controlled by the user, or on a service chosen independently by the user”. The foundation adds that “a default setting that silently redirects to a single provider is not a choice”.
Documents, it says, are “a user’s asset and, in many implementations, is also legally protected material: medical records, case files, tender documents and student data”. Under the LibreOffice AI policy, any transmission “must be the result of a user’s decision and not behaviour carried out in the background”.
No telemetry and no lock-in
LibreOffice already collects no usage data, and the third condition simply extends that rule to AI. The fourth targets lock-in: “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.”
Format and the right to say no
The fifth condition keeps generated text inside the Open Document Format, so an assistant cannot become a route back into proprietary files. The sixth is the one administrators will care about most: AI must be removable and “absent from the user interface” for organisations whose policy forbids it, even across thousands of workstations.
Why no integration passes today
Measured against the LibreOffice AI policy’s list, the foundation says, “there is no integration that meets all the requirements and that could be deployed, enabled by default, and supported throughout the entire lifecycle of a release”. Most commercial assistants fail the first and fourth conditions on their own, because they are built around one vendor’s cloud.
Why "No AI" Became a Selling Point
The foundation is frank that its position “has almost nothing to do with the technology”. It is about who pays and how.
The business-model argument
“For a company that sells subscriptions,” the post says, “an AI assistant justifies a price increase and strengthens the case for keeping all documents within its own infrastructure.” The Document Foundation, by contrast, has “no subscription tiers to protect, no upsells, no data to monetise”. That, it says, lets it take time to decide whether something is genuinely useful, “as we are not driven by quarterly results”. That is the economic logic of the LibreOffice AI policy in one line.
What AI added to office bills
The price history supports the LibreOffice AI policy’s argument. In January 2025 Microsoft and Google both raised consumer or business prices while folding their assistants into the standard plans. The table shows the published US list changes.
| Plan | Before | After | Change | AI included |
|---|---|---|---|---|
| Microsoft 365 Personal, per month | $6.99 | $9.99 | +$3.00 (+43%) | Copilot, 60 AI credits a month |
| Microsoft 365 Family, per month | $9.99 | $12.99 | +$3.00 (+30%) | Copilot |
| Google Workspace Business Starter, per user per month | $6 | $7 | +$1 (+17%) | Gemini |
| Google Workspace Business Standard, per user per month | $12 | $14 | +$2 (+17%) | Gemini |
| Google Workspace Business Plus, per user per month | $18 | $22 | +$4 (+22%) | Gemini |
| LibreOffice | $0 | $0 | None | None by default |
Percentages are the increase divided by the old price: 3 divided by 6.99 is 43%, and 4 divided by 18 is 22%. Existing Microsoft subscribers could switch to “Classic” plans without the new AI features, but only at renewal.
Bars are scaled to $264, so each width is the yearly cost divided by 264. LibreOffice’s bar is a sliver only so that it shows.
AI fatigue
TechCrunch tied the LibreOffice AI policy to a wider backlash against software packed with “unnecessary and bloated features, like chatbots and AI assistants”, which has pushed some developers to add switches that remove AI altogether. Browsers have moved the other way: Firefox now offers a Smart Window powered by Mistral models in beta. LibreOffice is betting that a meaningful share of users want the opposite.
The Download Numbers Behind the LibreOffice AI Policy
Does the LibreOffice AI policy actually attract users? The download figures are the best evidence available, and they are strong, though they cannot prove cause and effect.
A record first week
LibreOffice 26.8 passed one million downloads in its first week. Cybernews reported the exact figure the project gave, 1,031,162, and quoted the project: “This is the highest number of downloads in the first week of any LibreOffice release!” The count covers only the official download page, so it leaves out most Linux users, who get LibreOffice from their distribution.
Four million in September
The momentum held. On 6 October the foundation asked: “Why did LibreOffice 26.8 get over 4 million downloads in September?” Its own answer came in a conference talk by Mike Saunders at LinuxDays in Prague. Italo Vignoli, a founding member of the foundation, told Cybernews it was “the first time in LibreOffice history” with two consecutive weeks above one million downloads.
Bars are scaled to the 2026 figure, so 2025’s width is 24,830,220 divided by 29,755,011, or 83.4%. Figures are those Italo Vignoli gave Cybernews in September.
What the numbers do and do not show
By early September, 2026 downloads were already 4,924,791 ahead of the whole of 2025 (29,755,011 minus 24,830,220), a lead of about 20%. The weekly trend rose too: Cybernews, reading the project’s metrics page, saw 718,211 downloads in week 35 and 966,458 in the week ending 6 September, a rise of about 35%.
Vignoli linked the twin million-download weeks partly to back-to-back release announcements, and said “downloads in 2026 are higher than in the past” in general. So the LibreOffice AI policy is one plausible driver among several, alongside the European push to cut dependence on US suppliers.
What LibreOffice 26.8 Ships Instead of AI
The release that first carried the LibreOffice AI policy is not a thin one. The foundation put its effort into typography, writing systems and file exchange.
Better-looking text
Writer gains the Paragraph Composer, which balances word spacing across a whole paragraph instead of optimising each line in isolation, removing the alternating crowded and sparse lines of justified text. OpenType variable fonts are supported natively, and baseline grids now work inside frames.
More of the world’s writing systems
The largest body of work is for right-to-left and complex scripts. Writer detects paragraph direction automatically, Math gains named functions for the N’Ko and Adlam scripts used for Manding languages and Fulani, and new documents default to “start” rather than “left” alignment. The foundation notes that “none of this work has a commercial market”.
Safer exchange with Microsoft files
LibreOffice 26.8 preserves the newer Office chart types, such as waterfall and treemap, on a round trip, even though it cannot draw them yet. The follow-up 26.8.1 release on 2 October added around 40 fixes, mostly for DOCX, XLSX and PPTX files. Calc also gains Calculated Fields in pivot tables.
A donation banner, not a subscription
26.8 is the first release to show a periodic donation banner in the Start Centre. The foundation says LibreOffice carries “no advertising, no telemetry, no subscription, no account requirement”, and is funded by donations and ecosystem companies. In September, Zerodha’s FLOSS/fund gave $100,000.
How to Use AI With LibreOffice Anyway
The LibreOffice AI policy does not stop anyone using AI. It moves the decision from the vendor to the user or the administrator.
Extensions that talk to local models
The foundation says users “can install one of the available extensions”, found on the LibreOffice extensions site or elsewhere. “Most connect to a locally running model via Ollama, LM Studio or another OpenAI-compatible endpoint, which means the document never leaves the computer.” Our local LLM hardware guide covers what a PC needs to run those models well.
Vet every extension like third-party code
These extensions are third-party projects “at various stages of maturity”. The foundation tells administrators to check “whether a cloud endpoint is configured and what the provider does with the data it receives”. An extension pointed at a public cloud API quietly undoes the privacy benefit of the LibreOffice AI policy.
Local versus cloud, side by side
The table shows how each common set-up fits the conditions of the LibreOffice AI policy, and where the risk moves once AI is added.
| Set-up | Where the document goes | Main risk |
|---|---|---|
| LibreOffice default install | Nowhere: no AI, no network needed | Users paste text into web chatbots instead |
| Extension plus local model | Stays on the PC | Extension quality and hardware cost |
| Extension plus company-hosted model | Your own servers | Running and securing the model |
| Extension plus public cloud API | The provider’s servers | Retention and training terms you must check |
Privacy, Audits and the LibreOffice AI Policy
The strongest argument for the LibreOffice AI policy is not cost. It is accountability for where documents go.
“The only assurance that survives an audit”
The 26.8 announcement put it bluntly: “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.” For firms subject to UK GDPR, that is a cleaner story than a vendor’s assurances about AI processing. Our guide to private AI for UK businesses sets out the alternatives.
Digital sovereignty in the public sector
LibreOffice was already gaining ground in government before the LibreOffice AI policy was written down. In June 2025 Denmark’s Ministry of Digital Affairs announced it would move its staff from Microsoft Office to LibreOffice, following the German state of Schleswig-Holstein, which is migrating about 30,000 government workstations to open-source software. The foundation says Germany, NATO and the European Commission mandate ODF for public documents. We covered the wider trend in post-Brexit digital sovereignty.
Telemetry and shadow AI
A suite with no telemetry also sends no prompts anywhere. That helps compliance, but it does not stop staff pasting text into a free chatbot in the browser. The LibreOffice AI policy at the software level still needs a written AI policy at the company level to work, and the two should say the same thing about which data may leave the building.
What It Means for UK Businesses
For most firms, the question is not whether LibreOffice is “better” than Microsoft 365. It is whether the LibreOffice AI policy fits particular teams.
Who should take a look
Law firms, healthcare providers, schools, charities and public bodies handle exactly the documents the foundation lists. For them, the LibreOffice AI policy is a real control, not a slogan: a suite that cannot send content anywhere by default. It is also free, which matters for organisations reviewing a 17% to 43% AI-driven price rise.
Who will miss AI
Teams that already rely on Copilot or Gemini for drafting, summaries or spreadsheet formulas will notice the gap. A local-model extension can cover simple drafting, but not the deep links into email and calendars that the big suites offer. Mixed estates, with LibreOffice for sensitive work, are a realistic middle path.
How to explain it to staff
A LibreOffice AI policy decision is easy to misread as “IT has banned AI”. It is better framed as “AI only where we have checked where the data goes”. Tell staff which approved tools exist, which documents may never go into a cloud chatbot, and how to request an AI extension. A short written rule, reviewed every six months, does more than any software setting.
A practical rollout checklist
Start with a pilot group and a list of documents that must open perfectly. Note the dates: the 26.2 branch reaches end of life on 30 November 2026, so new deployments should standardise on 26.8. Large organisations should consider the enterprise editions sold by LibreOffice ecosystem companies, which add support contracts and three to five years of security updates. Our AI strategy service can help write the AI usage rules that sit alongside it, and our data protection team can map where documents travel today.
What Could Change the LibreOffice AI Policy
The foundation has said what would change the LibreOffice AI policy, which is unusual and useful.
On-device inference that just works
The post notes that “on-premises inference is becoming manageable on standard hardware” and that open models “are improving much faster than most of us had anticipated”. A small, capable open model that runs on an ordinary laptop, behind an open interface, would satisfy most of the six conditions at once.
Open interfaces and ODF output
The remaining hurdles are the provider-neutral interface and ODF-native output. If a standard interface for local models matures, the foundation could ship it switched off but present. Until then, the LibreOffice AI policy holds: “we will closely monitor where the technology is actually heading.”
LibreOffice AI Policy: Frequently Asked Questions
Does LibreOffice have any AI features?
No. LibreOffice 26.8 contains no generative AI features in its default installation, and no part of the suite needs a network connection to work. Users can add AI through third-party extensions if they choose. That is the LibreOffice AI policy in short.
Why did LibreOffice decide against AI?
The Document Foundation says no current integration meets its six conditions: user-chosen inference, no unauthorised uploads, no telemetry, no single provider, ODF output and full opt-out. It also says, as a non-profit, it has no subscription to justify.
Is the LibreOffice AI policy permanent?
No. The foundation calls it an assessment of today’s technology and says it will evaluate “very carefully” any approach that meets all the conditions.
Can I use ChatGPT or a local model with LibreOffice?
Yes, through extensions. Most connect to local models via Ollama or LM Studio. Check whether an extension sends text to a cloud service before installing it.
Is LibreOffice free for businesses?
Yes. It is released under the Mozilla Public License v2 and can be deployed without charge. Paid enterprise editions with support are available from ecosystem companies.
References
LibreOffice says ‘no AI’ is now a software feature (TechCrunch)
Yes, no AI is now a feature (The Document Foundation)
LibreOffice 26.8 announcement (The Document Foundation)
LibreOffice 26.8.1 improves compatibility and text import (The Document Foundation)
Where is LibreOffice going? (The Document Foundation)
LibreOffice downloads top 2 million as No AI stance wins users (Cybernews)
Microsoft follows Google with price bump for AI bundles (PCWorld)
Google Workspace price increase 2025 (9to5Google)
Denmark’s Digital Ministry shifts from Microsoft to LibreOffice (Computing)
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
