I disagree with your reading of the proposals. My interpretation was that only proposals (1) and (7) have severe limits on LLM use. Each of the other proposals allows LLM use with some very minor concessions or no concessions at all. It felt to me as though they were all worded very sneakily to try to attract votes from anti-AI developers while actually being very permissive, by waffling about the harms of AI in the preface before completely dropping the ball in the requirements section.
Removing all of the waffle, the proposals can be boiled down to:
Completely outlaws all use of LLMs in Debian contributions, whether code or documentation or communication etc.
Allows any and all LLM use provided that:
the tools are not legally problematic
the LLM output does not violate another copyright
the contributor assumes full responsibility
they’re not used for bulk operations without prior discussion
private data is not sent to cloud services
Allows LLM use for everything except for human-to-human communication. Also permits individual maintainers to ban LLM contributions to the project that they maintain (implying that maintainers may also permit LLM contributions to the project they maintain).
This one is particularly pernicious since it also “requests” (as in, a non-binding encouragement) that nobody uses LLMs, whilst acknowledging that for decision makers this “will involve uncomfortable compromises.”
Allows any and all LLM use provided that the contributor takes full responsibility, tags the contribution as LLM-assisted, and doesn’t use cloud services for private data.
Says absolutely nothing of substance. 8 paragraphs of waffle to place no restrictions on anything but to say that contributors are already bound by the Debian Social Contract and existing rules, and that they should be held responsible for their contributions accordingly.
Exactly as (5), but this time 9 paragraphs of waffle to say nothing of substance.
“In order to safeguard the project’s mutual understanding, trust and respect, we disallow the output of generative AI as direct contributions to Debian.”
12 paragraphs of hand wringing about the climate boils down to: “this text is just a position statement. This said however, we appreciate the disclosure of LLM usage.”
So by my reading, that makes:
Bans AI: (1), (7)
Allows AI: (2), (4), (5), (6), (8)
Allows AI with a restriction: (3)
Now read the titles of these proposals and their introductory preamble and ask yourself whether they are representative of the actual rules they will put into effect.
I disagree with your reading of the proposals. My interpretation was that only proposals (1) and (7) have severe limits on LLM use. Each of the other proposals allows LLM use with some very minor concessions or no concessions at all. It felt to me as though they were all worded very sneakily to try to attract votes from anti-AI developers while actually being very permissive, by waffling about the harms of AI in the preface before completely dropping the ball in the requirements section.
Removing all of the waffle, the proposals can be boiled down to:
So by my reading, that makes:
Now read the titles of these proposals and their introductory preamble and ask yourself whether they are representative of the actual rules they will put into effect.
https://lists.debian.org/debian-vote/2026/08/msg00352.html