Join Nostr
2026-02-28 15:14:30 UTC
in reply to

Ramin Honary on Nostr: Oh, OK, talking more on the generated output being public domain, regardless of how ...

Oh, OK, talking more on the generated output being public domain, regardless of how the training data was obtained.

I don't know why this isn't a bigger issue, as you say. Companies which sell software on the assumption that they own the code can now no longer assert that their code is proprietary. They may still be able to treat it as a trade secret though. If the code leaks, it becomes public domain unless they can show the code was not written by AI.

I am guessing the reason we don't see this being challenged in courts is that it would give reason for lots of companies to not use LLMs for development, and this could be detrimental to companies hoping to selll more LLM products. But then, none of these LLM companies are profitable anyway, so I don't know.