glibc AI policy

[Andrew Pinski]

On Thu, Apr 16, 2026 at 9:50 AM Andrew Pinski <pinskia@gmail.com> wrote:
>
> On Thu, Apr 16, 2026 at 9:27 AM Florian Weimer <fweimer@redhat.com> wrote:
> >
> > * DJ Delorie:
> >
> > > However, proving a negative is difficult.  In this case, as difficult as
> > > proving you have clear copyright status.  Making "prove you're not AI" a
> > > CoC violation opens up the opportunity to use this CoC rule as a weapon
> > > against contributors, and I oppose that.
> >
> > Yes, that's my concern as well.  It's not only that it might get
> > weaponized.  It really sucks if you are accused of being an LLM.  I
> > don't want reviewers doing that routinely.  It can easily lead to a very
> > bad first patch experience.
>
> I think it is a legit question to ask (once and if it seems obvious to
> the reviewer and others) if the used an LLM to help write the patch.
> It is definitely not legit to keep on asking for sure unless the
> person has admitted it in the past and failed to disclosure it.
> It is in a similar fashion to be asked if you can submit the patch for
> copyright reasons. Or if they copied it from somewhere else.
> Now if the reviewer is doing the asking all the time to every new
> person coming in (or a pattern of them asking when there is some bias
> going on); then yes there is a CoC issue.

The phrasing of the question does matter too and yes that can be a CoC issue.

Like an example of decent phrasing here should be something like:
Please make sure read you read our AI policy and if you used LLMs to
help produce the patch, please add "..." to commit message and please
read our contributor guidelines at .....

Bad phrasing:
I think you are using a LLM and our AI policy says we can't accept
this patch. Are you writing this patch with LLMs?

Thanks,
Andrew

>
> The reason why I mentioned (a pattern of) not disclosuring the use of
> LLM is a CoC issue is because it is conduct issue that has not been
> corrected; in the same way of not providing a clear copyrightablity
> issue. I am not saying one event of not disclosuring LLM usage is a
> CoC issue but if there is a pattern of it.
>
> Thanks,
> Andrea
>
> >
> > Thanks,
> > Florian
> >