LLM usage in Debian - General Resolution

[hydn (Hayden James)]

HI guys, so Debian is currently holding a General Resolution vote (their second of 2026) regarding the use of large language models (LLMs) and generative AI within the Debian project.

The vote explores multiple proposals on how to handle AI-generated code and content, ranging from strictly forbidding AI-assisted contributions in official Debian resources (due to copyright, quality, and ethical concerns) to allowing them under specific community guidelines.

It’s a really interesting look into how major open-source projects are navigating the rise of AI.

Check out the full text of the proposals here:

General Resolution: LLM usage in Debian

We also held a similar discussion which resulted in this addition to our code of conduct:

Use of AI (Artificial Intellegence)

We’re not anti-AI here. Used well, it’s a helpful tool, and many of us use it that way. But this community runs on people sharing their own knowledge and experience, so a few principles apply:

• Using AI as an assistant, for spelling, grammar, phrasing, formatting, language translation, or research, is acceptable and normal.
• If AI helped produce what you’re posting, especially AI-assisted search results, you must be upfront about it and present it as a quotation or reference rather than your own words.
• Do not post AI-generated content as if it were your own work. We’d rather read a human reply with a few rough edges than a polished block of text that isn’t really yours, and passing AI off as your own undermines the Community part of what we are.

The aim isn’t to police how anyone writes. It’s to keep this a place where you can trust you’re talking with real people sharing their own knowledge and experience, not reading something you could have generated yourself.

[Halano]

I believe Any modern open source project
should be like the kernel use AI but you should test it verify it before committing it

AI Coding Assistants — The Linux Kernel documentation

[Brian_Masinick]

The other day the General Resolution E was added in to reflect the current Debian LLM policy that’s been actively utilized to date. Until the AI landscape matures further, I personally believe that Proposal E allows AI technology to be used while ensuring that human validation is an integral component in the LLM usage in Debian.