LibreOffice Rules Out Built-In AI for the Foreseeable Future

Most software makers are racing to put AI into every menu and toolbar. LibreOffice is doing the opposite. The Document Foundation, the nonprofit that develops the open source office suite, has said it "will not add" AI to the software for the foreseeable future.

The statement came in a blog post this week. It follows the latest LibreOffice release in late August, when the organization had already noted that the suite "contains no generative AI features." The new post goes further. It explains the reasoning, sets out the conditions under which that could change, and makes clear that this is a position on today's technology, not a blanket rejection of AI.

A design decision, not an oversight

The Document Foundation calls the absence of AI a "deliberate design position." It rests on two commitments. Users' documents should never be uploaded to a server for processing, and no part of the software should need a network connection to work.

The organization says tens of millions of people and organizations use LibreOffice, and that those users should stay in control of their own data. Its argument focuses on institutions that handle sensitive material:

"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," the foundation wrote.

This points to a real weakness in many AI tools. They often require user data to be sent to a remote server and processed there. Once the data leaves the device, the user has no say in how it is stored or for how long. For a law firm, a hospital, or a public agency, that is a compliance problem, not just a matter of preference.

What would have to change

The foundation did not close the door entirely. It says no AI integration currently available meets all of its requirements. Two of the criteria it names:

1. Data stays local. User content must remain on the user's device.
2. No single-vendor dependence. The software must not rely on one AI provider.

The foundation says it has other considerations as well. It described its criteria as not a "definitive rejection," and framed the decision as an assessment of where the technology stands now, rather than a "judgement on the value of the sector."

The extension route

People who want AI in their documents are not locked out. LibreOffice says users can install extensions that connect the editor to AI models running locally on their own hardware. The default installation stays AI-free, and adding AI becomes an explicit, opt-in choice.

"Until then, there will be no AI of any kind in the default installation; extensions will be available for those who wish to integrate AI features, and we will closely monitor where the technology is actually heading," the foundation wrote.

Against the grain

The move stands out against the wider software industry. Many vendors keep adding AI features, sometimes at the expense of the user experience. Critics often describe this pattern with the term "enshittification": products degrading as they fill up with unnecessary, bloated features such as chatbots and assistants that nobody asked for. The backlash has been strong enough that some developers have built tools or settings specifically to block or remove AI features.

LibreOffice also belongs to a broader group of open source and alternative software that people use to reduce their dependence on Silicon Valley's largest tech companies. Keeping AI out of the default install fits that role.

Our Take

The core of LibreOffice's argument is about audits rather than AI. The foundation is not saying generative models are useless. It is saying that the only data-handling promise that holds up under scrutiny is "the data never left the machine." That is a narrow, testable claim, and it is hard to argue with. Recent incidents, such as agents leaking company screenshots, show how quickly that promise can break once data moves to systems outside the user's control.

The approach also mirrors a pattern seen elsewhere in enterprise software. Vendors serving regulated customers have been moving toward self-hosted, air-gapped setups. LibreOffice is reaching a similar endpoint from the other side: no AI by default, local AI by choice. For organizations that handle privileged or personal data, this suggests the real question is less "AI or no AI" and more "where does inference run, and who can prove it?"

There is a cost. Users who expect built-in drafting or summarization will need to find, install, and maintain extensions themselves, and the experience may be rougher than in suites that ship AI out of the box. Some may switch for that reason alone.

It is worth watching whether local models become capable enough, and easy enough to set up, that an integration meeting the foundation's criteria appears. The foundation has effectively published its conditions. If an extension can keep data on the device and avoid lock-in to a single provider, the debate could shift from principle to implementation. Until then, LibreOffice has turned "no AI" into a selling point, and other vendors serving privacy-sensitive users may take note.