Why is the focus exclusively on "copyright status"? Codeberg is meant to be the ethical alternative
@hsza
hi! i'm not on the codeberg team at all, but i wanted to reply my thoughts. The short answer is because, in meme terms "we live in a society". Copyright is an unavoidable fact and LLMs are right now working in a legal gray area which can be problematic, SPECIALLY for open source projects! What if suddenly the AI companies lobby laws to make sure the copyright of LLM-coded projects goes towards them? Or something like asking for percentages of profit, exclusive usage, etc. Even if you were to make your case via the judicial route, how many Open Source projects have the money to fight that? Working in a legal gray area is pracitcally a blank check that allows these companies to force projects to god-knows-what in the future.
also, "mostly" is up to too-lenient interpretation
To the rest of everyone here, I totally agree and am also worried about this! I've been thinking that the best scenario for llms is basically assisted typing. While the worst case is agentic hands-off vibecoding. The bad bit is this is a spectrum definitely, so where do we draw the line? I think for starters LLM-automated commits or PRs could be prohibited. The Godot team is prohibiting the usage of LLMs for human-to-human communication (as well as just generally in the project basically) which I think it's a good idea. I'm quite interested to see how this discussion evolves and particularly where do we draw the line? as I like seeing the possibility of good usages of LLMs. But again, the legal concerns are real and unavoidable
> Why is the focus exclusively on "copyright status"? Codeberg is meant to be the ethical alternative @hsza hi! i'm not on the codeberg team at all, but i wanted to reply my thoughts. The short answer is because, in meme terms "we live in a society". Copyright is an unavoidable fact and LLMs are right now working in a legal gray area which can be problematic, SPECIALLY for open source projects! What if suddenly the AI companies lobby laws to make sure the copyright of LLM-coded projects goes towards them? Or something like asking for percentages of profit, exclusive usage, etc. Even if you were to make your case via the judicial route, how many Open Source projects have the money to fight that? Working in a legal gray area is pracitcally a blank check that allows these companies to force projects to god-knows-what in the future. > also, "mostly" is up to too-lenient interpretation To the rest of everyone here, I totally agree and am also worried about this! I've been thinking that the best scenario for llms is basically assisted typing. While the worst case is agentic hands-off vibecoding. The bad bit is this is a spectrum definitely, so where do we draw the line? I think for starters LLM-automated commits or PRs could be prohibited. The Godot team is prohibiting the usage of LLMs for human-to-human communication (as well as just generally in the project basically) which I think it's a good idea. I'm quite interested to see how this discussion evolves and particularly **where do we draw the line?** as I like seeing the possibility of good usages of LLMs. But again, the legal concerns are real and unavoidable