LibreOffice promo graphic reading the private office suite with no forced AI, with a crossed out robot holding a folder — LibreOffice's own campaign graphic: no built-in AI, no sharing of personal data, no training of AI models. Image: The Document Foundation.

Every software update I install lately seems to add a chatbot I never asked for. So it caught my eye when LibreOffice went the other way. The free office suite says it will not add AI to its default install any time soon.

TechCrunch's Zack Whittaker reported on October 6, 2026 that LibreOffice is treating "no AI" as a software feature. I also read The Document Foundation's own posts on the topic. Here is what was said, why it matters for client data, and how agencies in India and the USA can use this news in real projects.

What LibreOffice Actually Announced

LibreOffice is made by The Document Foundation, a nonprofit based in Berlin. The suite includes Writer, Calc, Impress and Draw, and it is free and open source.

In late August 2026, the foundation released LibreOffice 26.8. The release notes said the suite "contains no generative AI features." The foundation called this "a deliberate design position rather than a missing feature."

A reader then asked a simple question under the release post: so, no AI is now a feature? The foundation answered with a full blog post titled "Yes, no AI is now a feature". Its main points:

LibreOffice does not reject AI out of hand.
It will not add AI to the default setup until AI can meet the project's own conditions.
Right now, no integration meets all of those conditions.
Users who want AI can install extensions, most of which connect to a model running on their own computer.

The foundation summed it up plainly: "Until then, there will be no AI of any kind in the default installation."

The Six Conditions LibreOffice Set for Any AI Feature

This is the part I found most useful. The foundation did not just say no. It wrote down what an acceptable AI feature would look like. Any AI in LibreOffice would need:

User-controlled execution. The user picks where the model runs: on their own computer, on servers they control, or with a provider they choose. A default that quietly sends everything to one provider does not count as a choice.
No content leaving the computer without permission. Sending a document anywhere must be the user's decision, not something the app does in the background.
No telemetry of any kind. LibreOffice does not collect usage data, and AI features would follow the same rule.
No dependence on a single provider. The foundation says an integration that only works with one company's API is a lock-in tool.
No compromise on format. AI output must be saved in the open ODF format with its structure and styles intact.
Fully optional. It must be possible to install it, remove it and hide it, including for IT admins managing thousands of computers.

Read that list again and swap "LibreOffice" for "our client's app." It works almost word for word as a checklist for any AI feature you build.

Why LibreOffice Is Skipping Built-In AI

Data privacy comes first

The foundation points out that its users include schools, hospitals, public bodies and law firms. Their files can hold medical records, case files, tender documents and student data.

As TechCrunch quoted, the organisation said: "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."

That line stuck with me. Many cloud AI tools process documents on remote servers. The user often has little say over how long that data is kept.

A different business model

The foundation was honest about money too. It argued that for a company selling subscriptions, an AI assistant helps justify a price rise and keeps documents inside its own cloud.

The Document Foundation has no subscription tiers or upsells. It says this lets it take time to decide whether a feature truly helps users, without chasing quarterly results.

Part of a wider pushback

TechCrunch noted that LibreOffice's stance goes against the trend of packing AI into every product. It also pointed to a command-line tool, covered by Ars Technica, that removes Apple Intelligence from macOS 27. Some users clearly want an off switch.

LibreOffice Writer window showing a user guide chapter on styles with the Navigator panel open — LibreOffice Writer with the Navigator panel. The suite runs fully offline and needs no network connection to work. Image: LibreOffice, via TechCrunch.

What the LibreOffice Decision Means for Agencies

At Debugged Pro we build software for clients in India and the USA. Here is how I see this news affecting day-to-day agency work.

1. Clients will start asking "where does my data go?"

A popular open source tool is now marketing the absence of AI. That makes data flow a selling point, not a footnote. Expect more clients to ask how your AI features handle their files.

Have a clear one-page answer ready. Say which features use AI, where the model runs, what gets stored and for how long.

2. "AI off" should be a real setting

LibreOffice's last condition is that AI must be fully optional. That is a smart default for client products too. Give admins a way to switch AI features off for a whole team, not just per user.

3. Local AI models are now a practical option

The foundation says most LibreOffice AI extensions connect to a local model through tools like Ollama, LM Studio or another OpenAI-compatible endpoint. In that setup, the document never leaves the computer.

For clients in health, legal or finance, this can be the difference between "no AI allowed" and "AI approved." Open-weight models keep getting better, as we covered in our look at what Mistral's new open source AI model means for agencies.

4. Avoid single-vendor lock-in

Build AI features behind a simple interface so you can swap the model provider later. It costs a little more time up front. It saves a painful rewrite when prices, terms or rules change.

5. Treat extensions like any third-party code

The foundation warns that AI extensions are built by outside developers at different levels of maturity. It tells admins to check whether an extension sends data to a cloud endpoint and what that provider does with it. That is good advice for any plugin, package or AI tool your team installs.

LibreOffice for Businesses in India and the USA

Privacy rules are getting stricter on both sides of the world. In India, the Digital Personal Data Protection Act, 2023 sets duties for anyone processing personal data. In the USA, sector rules like HIPAA already shape how health data can be handled.

A tool that keeps documents on the device makes compliance conversations easier. Here is a quick comparison to share with clients:

Question	LibreOffice default install	Typical cloud AI assistant
Built-in generative AI?	No	Usually yes
Needs a network connection to work?	No	Yes, for AI features
Usage telemetry?	None, per the foundation	Varies by vendor
Account or subscription required?	No	Often
Can AI be added?	Yes, through optional extensions	Built in, sometimes hard to turn off

LibreOffice 26.8 runs on Windows, macOS and Linux in 120 languages, under the Mozilla Public License v2. The foundation says organisations deploying at scale should look at enterprise versions from companies in its ecosystem. Those come with service agreements and security updates for three to five years.

If you want a structured way to map these risks, the NIST Privacy Framework is a free and practical starting point.

LibreOffice 26.8 release banner listing tabbed interface improvements, a new Draft Mode in Writer and better scripting support — LibreOffice 26.8, released on August 26, 2026, shipped with no generative AI features by design. Image: The Document Foundation.

A Simple Checklist Before You Add AI to a Client Product

I now run every AI feature request through a few quick questions. They borrow heavily from LibreOffice's list:

Where does the model run? Local, client-owned server or third-party cloud?
What data is sent? Full documents, snippets or only metadata?
Is consent clear? Does the user trigger each send, or does it happen in the background?
Is there an admin kill switch? Can the client disable AI for everyone at once?
Can we change providers? Is the model behind an interface we control?
What is logged? Are prompts and outputs stored, and who can see them?

We wrote more about hidden data risks in why AI memory features are a growing privacy risk. It pairs well with this checklist.

What Happens Next

The foundation was clear that this is not a permanent ban. It said local inference is becoming manageable on standard hardware and open models are improving fast. If an approach meets all six conditions, it will evaluate it carefully.

My guess is that "AI with clear controls" will become a selling point, much like "no AI" is today. Teams that design for privacy now will be ready either way.

Key Takeaways

LibreOffice will ship with no AI of any kind in its default install for now, and LibreOffice 26.8 contains no generative AI features.
The Document Foundation set six conditions for any future AI feature, including local control, no telemetry and no single-vendor lock-in.
Users can still add AI through optional extensions, most of which use local models via Ollama or LM Studio.
For agencies, the lesson is to make data flow clear, offer an admin off switch and keep AI providers swappable.
Privacy-first design helps with India's DPDP Act and US sector rules alike.

Planning an AI feature and unsure how to keep client data safe? Our team offers free software development consulting to help you scope it the right way.

About the Author

Nitin Kumar is Lead Project Coordinator at Debugged Pro. He manages software and AI delivery for startups and growing businesses in India and the USA, from first scoping call to launch. He writes about practical ways teams can adopt new tools while keeping cost, quality and data under control.