On 23/07/26 at 14:16 +0200, Ansgar 🙀 wrote: > Hi, > > On Thu, 2026-07-23 at 13:50 +0200, Lucas Nussbaum wrote: > > On 23/07/26 at 13:21 +0200, Ansgar 🙀 wrote: > > > Hi, > > > > > > On Wed, 2026-07-22 at 21:31 +0200, Lucas Nussbaum wrote: > > > > 4. **Explicit Disclosure:** When a significant portion of the > > > > contribution > > > >   is taken from a tool without manual modification, contributors should > > > >   disclose the tool's use. This may be recorded using Git trailers, > > > > such as > > > >   `Generated-By:` or `Assisted-By:`. > > > [...] > > > > 6. **Community Courtesy:** To respect the preferences of project > > > > members who > > > >   wish to avoid AI-generated content, contributors should clearly label > > > >   such content in mailing list and bug discussions (e.g., by > > > > identifying > > > >   such content with a clear disclaimer or a machine-readable tag like > > > >   `[AI-Generated]`). > > > > > > > > > Given what happened with rsync recently, I would suggest not including > > > such requirements. > > > > I'm sorry, I don't understand what you mean. Could you elaborate? > > There was recently an online outrage over AI use in rsync which > included harassing the maintainer up to wishing for them to die. > > This was started because the commit messages included attribution to a > LLM service. Not including those attributions would have made it less > likely. > > I don't think we should require contributors to expose them to this > additional risk. So including such attributions should be a free > choice.
Interesting, thanks for bringing this up. So, the dilemma is: it is useful for us as a community to be transparent about the use of AI, but people external to the project might use that information for harassment. I don't want to make it sound like I'm not taking harassment seriously. But I wonder if we could give it a try and, if AI-related harassment becomes an issue in Debian (AFAIK it isn't at this point), evolve our policy later (which would not require a new GR). Alternatively, someone could of course propose a ballot option that does not have those two points, but my initial feeling is that transparency is sufficiently useful to give it a try. Lucas
