Good luck to them! It’d be interesting to see how they prove scraping. Like do you find something unique to your website & then prompt the model to give you just that? So you use the citation/reference features that link to your websites?
Knowing the slimeballs at OpenAI I’d wager they’d have covered their tracks.
EDIT To be clear, I’m not suggesting they deleted evidence, but they “laundered” the data via a public training dataset like Eluther AI’s “the pile”.
Oh no! 😨
Oh yeah 😈