Codeberg has decided to ban predominantly AI-generated repositories after a community vote
Codeberg has introduced a policy banning repositories that mostly consist of code generated by AI tools. The change follows community concerns about “vibe-coded” projects, where AI produces most or all of an application with limited direct human contribution. The proposal passed with 358 votes in favor and 144 against, representing about 71% support.
Codeberg has now added the restriction to section 7 of its Terms of Use, specifically naming tools such as Claude and OpenAI Codex. The platform cited unclear copyright and licensing status, along with the risk that largely AI-generated projects may include harmful or insecure code without adequate safeguards.
Repositories may still contain smaller amounts of AI-generated code, but the policy applies when it makes up most of a project. Human review, editing, or maintenance does not appear to create an exception. Codeberg has also not defined a technical threshold or explained how it will identify and enforce violations, contrasting with platforms such as GitHub that continue to expand their AI coding tools.
- Version Control System
- Free
- Open Source
External links
- Protecting our FLOSS commons from LLMsCodeberg News • Official source
- Codeberg Git Hosting Platform Bans Projects Mostly Written by Generative AILinuxiac
- Codeberg takes its side in the open-source scene's AI debate by banning vibe-coded projectsXDA
- Codeberg’s programmer user base overwhelmingly votes to ban slopcoded projects from its platformOSnews
- Codeberg members vote to ban AI-generated codeOMG! Ubuntu
Comments
This are great news, codeberg is focusing in projects made by humans mainly while not banning the use of AI for small tasks. The people who see an issue with this probably don't know how to write code directly, so they just want to generate all the vibe coding possible for their "apps". But codeberg is a non profit FOSS organization, and as that they value morals too, a thing that sadly big LLM doesn't meet as they are made by stole content of others that will never get anything from that. With this decision they are discouraging the use of their platform for making slop, if you want to make slop, please, still using github rather than wasting resources from codeberg.
Good W from
Codeberg, unlike the PISS (plagiarized information synthesis systems, 1, 2) that is Microslop GitHug, the serial GPL Violator.
TL;DR Slop Coding never work (1, 2, 3):
Speaking of PISS, take a listen to Dr. Andy Farnell's take on slop being plagiarism.
Even Cornelius Schumacher from
KDE Plasma is also in the No-Slop side, that you should invest in your OWN identity than pretending to be productive (slopping) (that is, return to the basics).
This leaves a lot up to interpretation and by galvanizing everyone behind a hateful cause is a straight road to making your community toxic. This provides Codeberg with the reasoning to get rid of truly junk content but innocent projects will also be boycotted off the platform.
I'll have to pay more attention to devs who use Codeberg, would be a sign of quality that your code doesn't violate the AI-slop restriction
Good thing we still have disroot's git
become inhumanly good or humanly competent.
A good decision.
Lol no. It just depends on how it's managed. If it's not malicious, why ban it?
Love it or hate it its the new way a lot of code is going to be written from here on out, quality will improve as models get better naturally. Even Linus Torvalds uses LLM's to code for some of his smaller projects.
A great decision, people complaining at your comment didn't even read the new. It doesn't say it will ban AI use, it says it will ban AI made projects. If you just use AI to support you at your coding and not as the main thing, no one will stop you from using codeberg. About the other comment about Linus torvalds, it is funny that you mention him because he even said that the PROJECTS should be able to choose how they want to deal with LLM...