They care when they get big enough to attract attention from people like state AGs who can actually put the hurt on a bit. Uber and AirBnB both hit this point years ago; OpenAI's starting to hit it.

Tons of websites and books they scraped had copyright notices.

If using data violating some ToS taints the model trained on that data, then all of OpenAI's models are tainted by the millions of ToS'es they broke.

But whats the remedy in that case? Being banned from the service maybe, but no court is going to force a "return" of the data, so DeepSeek can't use it. It's uncopyrightable.

As others have noted, if one company agrees to the ToS, asks "the right" questions and then publishes the ChatGPT answers, there is not violation of ToS. Then a second company scrapes the published Q&A, along with other information from the internet and again there is no violation (not more than the violations of OpenAI).

> OpenAI is saying that their service was used in violation of their TOS

Which is the most ridiculous argument they could use because they didn't respect any ToS (or copyright laws, for that matter) when scraping the whole web, books from Libgen and who knows what more.