[QGIS-Developer] Floating an idea: ban AI based contributed from non-core developers?

Wed Apr 1 13:57:41 PDT 2026

Hi,

Personally, I am -1 on this one, but not  because I think AI
contributions are good, quite the contrary. But because the cure
(banning AI LLM) could be worse than the disease.  Let me explain.

To me the situation is a bit similar to the alcohol prohibition law.
While the intent to reduce alcohol consumption was good, the
unintended consequences were the rise of organized crime.

If we ban outright AI contributions, people will not stop using them,
*they will stop disclaiming their use of AI tools*. We will end up
with a lot of "stealth" AI PR.  And I suspect spending much more time
chasing the liars.

Even in the long term, let's imagine the lawmakers decide AI generated
code is not compatible with GPL licence, it would be much easier to
scrape this code if each PRs with AI are correctly flagged today.

With our current AI tools policy at least people are not afraid to
disclaim their use of AI.

We also run the risk of mistaking a genuinely new contributor to an AI
agent parrot. It happened before with the Slop label, I don't see why
it would not happen with banning new AI contributions.

But I agree we need to do something about PR flooding, I actually like
the idea of Even:

> So, half joking/half serious, let's
> use AI tools to detect bad AI output ?!?

I would add to the suggestions:
* to detect IA accounts they usually have a "too pretty" github
profile, too many technologies, too many emojis etc..

* I would like to suggest one more than may seem unusual at first; It
would be to add more "good first issue" so we can funnel new
contributors energy to these lower risk issues but that are still
interesting to have in QGIS. And requires new contributors to complete
at least one while letting use time to sort which contributor is worth
taking reviewer time and which contributor is a LLM parrot.

The state of IA LLM is changing rapidly. I appreciate that we can have
this discussion and that we can periodically update our policy towards
AI.

But for now my position is to keep our disclaimer policy and wait and
see. I would be very curious about the outcome of the total ban on
other FOSS projects in a year from now.

Finally, concerning the label, I would like to note the label is not
named 'AI Slop' but 'Slop /!\/!\' which I think is worse and could be
seen as a personal attack on their works.

I got lucky to join the QGIS project before the advent of AI tools,
but I don't think I would have liked to be labeled as we do today for
new contributors.

So I second the suggestion raised earlier in the thread to rename the
label to something more neutral like "AI generated" or "AI Warning"

Kind regards,
Valentin

Le mer. 1 avr. 2026 à 15:04, Stefanos Natsis via QGIS-Developer
<qgis-developer at lists.osgeo.org> a écrit :
>
> Hi all,
>
> I'm on the same page with Nyall on this one. While not perfect, a no AI policy will improve the SnR of new PRs and that's important granted that we already struggle with a queue of ~100 PRs!
>
> > QGIS already has a problem with welcoming new contributors, and a policy giving more special rights to core contributors will only make the situation worse.
>
> I strongly disagree with this opinion, especially before the inflow of AI assisted PRs. It's actually quite the contrary, QGIS is so welcoming to new contributors that we are starting to face issues with AI contributions! Yes, sometimes PRs stay in the queue for too long, however reviews are helpful and not dismissive.
>
> It is sane and healthy to not treat core contributors or seasoned veterans the same as newcomers. Core contributors are by definition trusted so they can use the 'risky' tools. This is not gatekeeping in the same way that it is not gatekeeping to limit commit rights to core contributors only. We could still adapt this to requiring a minimum of let's say 50 merged PRs before AI tools are allowed.
>
> Of course, since this is not enforceable, practically anyone can still use AI tools to review his code, or even generate new code and we won't know, and that will be OK, as long as the code is good!
> Overall SnR will be improved though, as anyone wanting to contribute will need to put more effort into it.
>
> Best
> Stefanos
>
> On Wed, 1 Apr 2026 at 14:47, Greg Troxel via QGIS-Developer <qgis-developer at lists.osgeo.org> wrote:
>>
>> I should be clear that really I lean to "no AI contributions, at least
>> for now" - the legal situation is unsettled, as is the social situation,
>> and declining to engage while others figure it out seems best.
>>
>> What I really meant is that compared to where we are now, restricting
>> contributions to those with a track record and human relationships is a
>> positive incremental step.  However, I see the lack of equal treatment
>> as a a bug.
>>
>>
>> As for concern about new contributors and what amounts to "the kids
>> these days just want to vibe code and if we aren't ok with that they
>> won't play with us", I think that's ok to defer worrying about.  If
>> that's really how it is, the Free Software world has much bigger
>> problems well beyond the scope of any one project.
>>
>> _______________________________________________
>> QGIS-Developer mailing list
>> QGIS-Developer at lists.osgeo.org
>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
>
> _______________________________________________
> QGIS-Developer mailing list
> QGIS-Developer at lists.osgeo.org
> List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer