Skip to content

Comment on OpenAI and Google ask the government to let them train AI on contentparent

Comments

Copyright office recently ruled that outputs are generally not copyrightable without significant human changes. Prompts are also not copyrightable. I'm actually surprised that is not discussed around these parts more because it essentially means any app etc built by AI has no copyright protection. Thus code from llms cannot be opensourced using opensource licenses. Business cannot stop users from stealing their apps etc.

It also means that if you build something with a popular llm then big tech now has your inputs and outputs, and you have no IP to stop them from stealing either.

The next ruling from the copyright office is around fair use that big tech is pushing for.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.