Proposal Assembly 2026: ToU extension to prohibit LLM-extrusions #1253

Here is the proposed addition for the ToU that @Gusted and I agreed on post-assembly for the upcoming membership vote. 🙏

The discussion for this proposal is available for Codeberg e. V. members in the forum: https://forum.codeberg.org/d/139-resolutions-for-assembly-on-codeberg-taking-a-stand-against-ai

Signed-off-by: gedankenstuecke gedankenstuecke@noreply.codeberg.org

As highlighted by @Gusted for #1254, status quo for now due to being sent for voting, minor edits can be made by the board/presidium if the text generally passes.

"You must not share projects"

• Is this reffering to PR's cpmmitted to (other) Codeberg repositories
• Is 'sharing' including sharing code, outside Codeberg, from a Codeberg repository
• Is it accepted that such repository is kept 'private' (note, here sharing could also include allowing individuals access to said provate repository)

"that mostly consist of code "

• What is 'mostly' ?

Beyond it being too late for substantial changes, it's out of scope for this change. It uses the identical language around sharing as used for the other points on acceptable/unacceptable content.

If you understand what sharing means in the context of "You must not share any content that's expressing hate or encouraging violence", the same applies here.

I am in favor of this proposal. The copyright issues alone are sufficient to be problematic for a non-profit.

@infosec812 wrote in #1253 (comment):

I am in favor of this proposal. The copyright issues alone are sufficient to be problematic for a non-profit.

The copyright issue is always relevant, not directly related to this proposal alone.
It’s also quite often broken, as is LICENSE

Seems to me like it goes a bit far to not be allowed to share it for educational purposes (the readme could carry a prominent notice; the main file could be broken on purpose so it cannot be run as-is; etc.) as well as prohibiting manually reading the code and then sharing it. E.g. if it's a 150-line python utility, you can perfectly well read, understand, and test it, and only then share it, but this would now be disallowed without writing another 75 dummy lines. I am really conflicted what to vote on this specific wording for the terms of use since I also see the upsides

Seems to me like it goes a bit far to not be allowed to share it for educational purposes

Slapping the label "educational purposes" is not a get-out-of-jail-free card. It doesn't magically make the copyright issues go away. Even if it did in certain jurisdictions, it certainly doesn't in others.

if it's a 150-line python utility, you can perfectly well read, understand, and test it, and only then share it

As far as I understand it, this isn't about the wording being fool-proof & covering all corner cases. It's supposed to capture our intent & sentiment as well as reflect reality: even while you could read that one 150 line Python script & fully test & understand it, reality shows that regularly reviewing huge LLM contributing quickly degenerates into "LGTM, merge" due to fatigue etc. The majority of "mostly just LLM-generated" code will not be thoroughly understood and reviewed.

Don't let perfect be the enemy of good. Such wording doesn't need to cover every possible case.

I'm in favor.

Why is the focus exclusively on "copyright status"? Codeberg is meant to be the ethical alternative, so the ethical issues should also be brought up, not shoved aside.

also, "mostly" is up to too-lenient interpretation. would that mean a codebase that is 49% slop will be platformed? something like "in significant part" would be better. make it clear that no contributor notably outsourcing their thinking to an LLM is part of the target audience here.

"A contribution should be worth more to the project than the time it takes to review it."