LibreOffice's 'no AI' stance meets LLM-assisted code

The Document Foundation says LibreOffice has no built-in AI features and that core developers do not use large language models, while community contributors are allowed to use LLM coding assistants under disclosure and review rules. Three recent LibreOffice commits acknowledge such assistance, drawing criticism from some users who say the practice conflicts with the project's public no-AI message.

Commits that credited AI tools

A Fediverse post by a user with the handle argv[-1] highlighted three commits this year. Two April commits, 'Inheritance when setting paragraph style font size to 100% or +0pt' and 'Enable relative font sizes in Calc cell styles,' credit Claude with helping identify the fixes. A September commit, 'Add Copy Image button to screenshot dialog,' acknowledges using GitHub Copilot in VS Code to generate the BitmapTransferable class, while its copyButtonHandler method was written manually with LLM or code-completion assistance.

TDF's position and response

TDF has made the absence of built-in AI a selling point. It published a September blog post saying 'no AI is now a feature,' and in October announced 9.4 million downloads over eight weeks, excluding Windows updates and downloads of older releases. On Mastodon, TDF said no core developers use LLMs, that it requires contributors to disclose LLM use, generally discourages the practice, carefully reviews submissions, and prohibits LLMs for students.

Policy context and reactions

TDF spokesman Italo Vignoli pointed to the project's Development/AI policy and said there was no departure from public statements about LibreOffice 26.8 because those statements concerned user features, not development. Some users were unhappy; Argentinian developer Federico Schonborn compared the situation to a bakery announcing no asbestos in frosting while using it in filling. Commenter 'Skjeggtroll' said it seemed odd to ban all LLM use for commit messages while still allowing it for code.

Sources: The Register