Codeberg seems to ban vibecoded Projects; reason might be german copyright law
>It looks like Codeberg want only copyrighted material in their service, so it is reliable in the future that e.g. licenses must be followed (e.g. GPL), and copyright doesn’t suddenly get declared as being of the model owner, and it isn’t a copy of something else.
That is a cautious reasonable position - in early days of LLM coding (3 years ago!) indemnity from model companies was a major issue globally because of the lack of clarity of the law around this. The US specifically has settled on it being (effectively?) public domain. But I don’t think that is fully settled, and it certainly isn’t settled in international copyright law.
The goal of the vague “mostly” in the Codeberg change is to ensure there is enough human input to the code they host, to be reasonably sure under German copyright law it is copyright of the person sharing it.
edit: link to poll that caused it (might be down due to high traffic) https://codeberg.org/Codeberg/org/pulls/1253#issuecomment-19820434
edit 1: i dont defend/oppose this move, i just find it interesting
This is just to let their prejudice flag fly. There’s no way to actually tell AI was used if the planning and assisting meta-data files aren’t committed to the repo. Nothing has truly changed. They might as well ban anyone who thinks about the color pink more then twice per day - it would be just as easy to enforce.
You might not be able to tell, but the signs something is LLM generated are pretty obvious to more experienced programmers.
That, plus, it’s driving away vibe coders just because they don’t want to be there. The rule doesn’t have to be enforceable 100% in order to be effective.
Yeah, just like more experienced artists can see the AI generated pixels.
Are you trying to make some point? Because, yes they can.
I hate genAI, but this is dangerous. I’m a writer and I’ve had my stuff accused of being genAI when it’s not. And there was an incident several years ago on a prominent art subreddit where an artist was banned for posting AI art, and when the artist proved it wasn’t by sending layer files, the mod upheld the ban and told the artist they should change styles because it “looked” like genAI. It’s not a great feeling.
Yes, I am. I have a global pre-processor directive for all of my prompts, so it always includes “match existing code style, formatting and conventions.” The result is indistinguishable from hand written code. Thinking you can magically discern what was written by AI is delusional.
Sorry but you are clearly delusional. There is no magic involved, just being able to look at the structure of the code and the pattern of the commits. It’s fairly easy in fact.
Oh, I see. You assume agentic AI that can perform it’s own commits. That would make it easy to tell, but you’d have to be really stupid to give AI such unfettered access to your repository.
So 1) we absolutely know people are much stupider than that. and 2) doing exactly that has widely been observed.
Oh, without a doubt. They’ll get some low-hanging fruit, I’m sure.
And many, if not most, of the people vibe coding projects on git repos are that stupid. People who have never coded and never used anything other than
git clone. That is just a fact of many of the people who do it. They are the people who think they are tech geniuses because they once made an LED blink on arduino or made the millionth timer app.It is just a fact of who is flooding the most slop projects. There are plenty of people who use AI in subtler or more useful ways, but that is not what is the target of anti-vibecoding rules. It is the 1000 slop projects generated bloated repositories fully in 1 day that are killing bandwidth and storage limits for everyone else.
That makes sense.
All you need is another one of those global pre-processor directives! Then you’ll have commits that exactly match human output just like with the code itself!
Technically, yes.
“Make commits sporadically, using inexact terms, occasional typos and every once in a while, forget to include a file that then necessitates a follow-up commit of a single line that simply says ‘whoops’”
No need to assume that. Just by using LLMs to assist with code writing, the human in the loop changes their commit pattern.
“Suspicious commit pattern” sounds like a pretty flimsy basis for banning a human from the platform. If nothing else, this futile crusade will be entertaining for almost everyone; except for their support staff who will need to investigate and either defend or repeal the ban each time there’s a false-positive, and a person is accused of wrong-think because yesterday they commit twice in an hour whereas they usually only commit once per hour.