Hi, On 03/03/26 at 14:07 +0200, Jonathan Carter wrote: > On 2026/03/03 02:07, Thorsten Glaser wrote: > > > Just as in last year, I don't think the project as a whole is ready > > > for a vote on the GR. > > > > I donβt fully disagree. This is forced by the people actively > > trying to allow slop in. > > I don't agree with this sentiment. I don't want slop in Debian, but as more > and more upstreams increasingly use various AI tools, I believe it's > becoming increasingly urgent for Debian as a project to have some guidelines > on this. Many of us feel the same way. As Ansgar mentioned in another > follow-up, rsyslog should probably be out already. So policy would really > help in *preventing* slop and code with questionable copyright. > > That said, I don't personally think a GR is urgent right now, but > discussions about it is important so that we can find some good common > ground for when a GR on this does happen.
I initiated this particular thread mainly in response to various attacks against people using AI in the context of Debian (or against people were assumed by the attacker to use AI), both in public and in private, such as: - AI-shaming; asking for apologies from people assumed to use AI; calling AI users "AI shysters" - people arguing that using AI is a CoC violation and should result in people getting expulsed from the project - people saying that we should have a GR to ban AI contributions I initially felt that a GR was quite urgent given the frequent attacks we were seeing. However, the discussion has been generally civilized and interesting. My wild guess is that if we were to vote on a GR today, the winning option would probably be very nuanced, allowing AI but with a set of safeguards. So maybe, assuming the atmosphere remains calmer and productive, we don't need a GR after all, and we can continue exploring all the facets of this complex topic in calm mailing list discussions. Also, I believe that there are at least three different questions to address, and that it might be better to address them in separate discussions, or at least clearly identify them: A. What we should do about models in the archive (that was the main focus of last year's discussion) B. What we should do about upstream code that was (B.1.) AI-generated or (B.2.) written with AI-assistance C. What should we done about contributions to Debian directly (not upstream projects) that were AI-generated or written with AI-assistance. My GR draft focused on (C.) because that's what was attacked. Thorsten's is mainly about (A.), addressing (B.) and (C.) only in the paragraph quoted by Ansgar. Outside of this specific paragraph, I see no overlap between Thorsten's proposal and mine (we could decide that AI models are unacceptable in the Debian archive, while we allow contributions generated by those models). Lucas
