Disallow or discourage use of "AI" tools (Christoph Cullmann)

[Christoph Cullmann]

Hi,


On Sunday, May 18th, 2025 at 09:00, Justin Zobel <justin at 1707.io> wrote:

> On 17/05/2025 01:40, Christoph Cullmann wrote:
> 

> > Hi,
> > 

> > just as a concrete example: what to do with
> > 

> > https://invent.kde.org/frameworks/syntax-highlighting/-/merge_requests/698
> > 

> > That is no AI spam but something that doesn't look broken and the submitter did
> > do manual work.
> > 

> > Can I now accept that just as MIT?
> > 

> > Greetings
> > Christoph
> 

> If the contributor cannot tell you the license(s) of the code that was used to generate the code, then it's literally gambling that this code wasn't taken from another project by Gemini and used without their permission or used in a way that violates the license and opens up the KDE e.V. to litigation.
> 

> This is an absolutely possible scenario if the author happens to look around for their code being re-used. KDE e.V. CAN NOT accept AI contributions because the source of the code isn't known.
> 

> It really scares me that we would even consider accepting this. I fully understand that it is impossible to tell if a user is lying about generating code with an AI, but we have to at least remove the KDE e.V. from possible harm by rejecting code unless it is sourced from a license and privacy respecting model. Which I'm sure there are very few of and the ones that exist would have very little code as every single piece of code would have to be audited by the owner of the model to ensure that it can be distributed and used in their software, and that the owners accepts this.
> 

> Of course, this still all boils down to trusting contributors. They can get code from anywhere and claim at as their own. AI just makes it much easier for them to do it.
> 

> Justin

there is now input about what was put into the model to get the result, now the question is what to do.

As one input was an LGPL file and some stuff with MIT, the stuff would be the LGPL I would assume.

Naturally the question is what to say about the model itself, but if that implies that all people are out that use the today
often integrated AI stuff for help, that is not that helpful.

Greetings
Christoph
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-devel/attachments/20250518/a08117b8/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 343 bytes
Desc: OpenPGP digital signature
URL: <http://mail.kde.org/pipermail/kde-devel/attachments/20250518/a08117b8/attachment.sig>