Said and done - just as with Github I now only have forks on Codeberg. Self hosting is always the best protection against purity testers.

https://git.sync.wtf from now on.

@troed @Codeberg If that's true, I'll stop keeping (almost) silence because registration requires an inaccessible CAPTCHA, I'll start actively recommending against it.

@pvagner @Codeberg @troed Sorry but "a friendly support" as an accessible CAPTCHA alternative is absolutely not an option in 2026. there are audio captchas, math captchas, checkbox based captchas, captchas where you have to pick the correct sound (like "Select which of the three is a cat meowing"), something I definitely forgot — and "contact support to register" is the most humiliating and discriminating option you can ever imagine. It implies human intervention, so it's not #Accessibility, it's a crutch. And I'm not even talking about LLM here.

@pvagner Sorry again, but I don't care about anything except for #Accessibility. I'm blind and I spend too much time fighting inaccessible solutions.

@menelion @pvagner This is the problem with choosing platforms on moral grounds instead of practical grounds. Platforms which consider themselves moral stewards tend to keep making moral decisions, and it's almost inevitable that they'll eventually make one that you heavily disagree with.

@miki @pvagner Earlier I was resenting something like culpability because I had to tear myself in two between moral and practical. Now I'm absolutely 150% calm: I'm deprived of a basic sense, I need special requirements for me to work and live my full life, so #Accessibility is the only thing I care of. If a company preaches so-called "inclusion" and "diversity" and is super aware of people's pronouns, allows to select a gender from a list of fourteen items, has a carbon footprint on each page and writes long posts about why AI and "big tech" are evil, but absolutely doesn't give a flying F about me and my needs, I don't care how moral this company is, sorry for being harsh.

@menelion It's *funny* how disabled people are almost always excluded from these diverse communities. Visible but ignored. @miki @pvagner

@miki

Platforms which consider themselves moral stewards tend to keep making moral decisions, and it's almost inevitable that they'll eventually make one that you heavily disagree with.

There are two problems with that:

• Moral decision doesn't have to be inaccessible or exclusive in other ways
• Unfortunately so called "moral" decisions are often just political and not based on real science or solid proofs.

First is simple: resources are limited and for developers of Forgejo it could not only be out of scope, but even out of their awareness.
Example of the second option is LLM usage rules some project introduce. There are a lot of controversy in this matter and despite some people claiming that it's "rather obvious that %insert_their_opinion_here%", every argument they'll give you (if they bother to) most likely will have opposing arguments. And they won't be interested in a productive discussion. They'll most likely just want to shove the thing they don't like under the rug and feel good about it.

And if this thread would be a bit more popular, I'd be ready to bet on "there's no moral AI usage" comment appearing under my comment which would've been a good proof of what I'm talking about.

P.S. A better example of trying to think about the issue is SFC recommendations:
https://sfconservancy.org/llm-gen-ai/llm-backed-generative-ai-recommendations.html
Although I think that they leaned a bit more into the "anti-AI" movement, they shown consideration for both sides and tried to achieve something productive and inclusive. This deserves respect.

@menelion @pvagner

@pvagner

These two items at the very bottom of the list meant that I removed all my (non-fork) presence on Codeberg yesterday since a few of my projects were directly affected:

* Projects written and maintained with heavy use of LLMs
* Projects heavily tied to the LLM ecosytem, e.g. LLM-written tools to ease LLM usage

It's pretty clear that the Codeberg ToU changes are simply a reflection of general AI-hatred with no actual insight into how LLM tools are used by FOSS maintainers nowadays behind it.

https://blog.codeberg.org/protecting-our-floss-commons-from-llms.html

@menelion @Codeberg

@menelion @troed llm is good when it comes to coding. tried new models. These can be really helpful.

@troed@swecyb.com @Codeberg@social.anoxinon.de
Apparently that is correct.

https://codeberg.org/Codeberg/org/pulls/1253

The rule is

7. You must not share projects that mostly consist of code written by "generative AI"-tools (including services such as Claude, OpenAI Codex). Such projects having an unclear copyright status (see requirements § 2 (1) 1 and § 2 (1) 3) and furthermore have little safeguards to ensure that they do not include harmful code (c.f. § 2 (1) 5).

@phil

Thanks - alright. That's just silly, _I_ take responsibility for the legal status and security of my own code but #Codeberg can of course institute whatever rules they wish. I think they'll be surprised at how many well known FOSS projects use LLM tools today.

I do already host my own Forgejo instance so for me personally leaving will be easy, but this is a blow to European open source where many projects will just have to leave for Github et al.

@Codeberg

@troed@swecyb.com @Codeberg@social.anoxinon.de
It is what it is. I don't think any organization seriously concerned for retaining ownership of their code is using a service provider for Git. It's not like setting up a server for it + runners is difficult or expensive (and frankly owning the actual infrastructure is just common sense).

@troed you sure need a lot of letters to spell "I am stupid".

@Codeberg

@walsonde You only needed one sentence. Congrats 🏆
@troed