If Google were specifically going to ChatGPT to get its output and train off of it, they could be sued for breach of contract - and OpenAI would likely have a pretty good argument:

- they specifically tried extracting and learning from our model when it says you can't in our TOS

- this makes it easier for them to compete with us via the data they obtain in their breach of contract

- more businesses and enterprises might pass up on renting a shared or dedicated instance from us if they can just get it from Google

> If Google were specifically going to ChatGPT to get its output and train off of it

But (correct me if I'm wrong) I don't think anyone anywhere is claiming that's what happened. The claim was just that Google looked into using existing chats that it scraped from another website.

Edit: realizing you're probably replying specifically to the question I asked, "and what even is the penalty of that TOS violation and how enforceable is it?" In which case, yeah, that's a decent clarification to add, sorry for pushing back on it.