Post #1966385

2026-05-03 18:15 UTC

We updated our contribution policy for @teamtype last week, which now states the following: &quot;We do not accept any contributions that contain content generated by LLMs. This also includes all written communication on GitHub. We will close any pull request, issue, or discussion that does not comply to this policy.&quot; In our team, we have different reasons for this, but myself, I have a few of them: 🧵

Replies (9)

@blinry@chaos.social 2026-05-03 18:16

First of all, I want to be proud of the things I create! This happens mostly when I work through hard problems, and then still solve them in the end. That means that the process of working on a project might be more important to me than the actual outcome. If we started outsourcing the work to LLM tools, the satisfaction of having new features would feel hollow.

Open ##1973012
@douginamug@mastodon.xyz 2026-05-03 18:44

@nick dunno if Karrot policies yet, but this is a nice thread.

Open ##1973042
@janl@narrativ.es 2026-05-03 18:44

@blinry 🧡

Open ##1973045
@hipsterelectron@circumstances.run 2026-05-03 19:14

@blinry @teamtype love you love your work thanks for this

Open ##1973046
@gedankenstuecke@scholar.social 2026-05-03 19:33

@blinry that's a really great step, glad to hear it! ❤️

Open ##1973047
@tanuva@chaos.social 2026-05-03 19:41

@blinry Thank you very much! I typically argue why LLMs are bad, but this is a much better spin. I‘m concerned about what LLM usage does to *me*. (And the project as a whole.) Perfectly on point!

Open ##1973048
@lumi@snug.moe 2026-05-03 19:43

@blinry @teamtype forge.starlightnet.work/Team/No-AI/issues/65 :)

Open ##1973049
@mislav@hachyderm.io 2026-05-03 20:02

@blinry I regularly contribute to a project on Codeberg (Readeck) that also has an anti-AI policy on code contributions, as well as being one of the only read-it-later solutions out there that boasts no LLM-powered reading features, and I find your rationale well-explained and completely reasonable. The matter of it not being legally clear whether LLM-synthesized output can abide by a copyleft license aside, I’ve found that I learned a lot while hand-crafting the code I contributed, and in that sense the process was more valuable than the destination.

Open ##1973050
@dj3ei@mastodon.radio 2026-05-04 11:42

Curious question: When a PR comes in, how would you know its author abided by your policy? Personally, as a retired dev, I have not touched AI and can afford to continue to not touch it. But I do have a few low-traffic FLOSS projects out there (which see a PR once a year or even less often). So that's the background to my question. @blinry @teamtype

Open ##1973051