I don't often confess this in public, but I am a contract lawyer....

There will be a time element to this; my understanding is that ChatGPT works from a copy of the internet that's two years out of date (this the info will be out of date also). Secondly, if the data is scrapable that suggests it's not paywalled or in a walled garden; if freely available it's likely to be in public domain. If not public domain per se, but registered as a patent for example, then the idea may be public, but protected legally by patent law. Trade secrets already have very strong extra legal protection in most jurisdictions.

Short answer is that I don't think anyone will have begun thinking about this in enough detail and we probably need some case law to set precedents, where people start suing GPT and Sam for plagiarism!