Yeah, maybe a spotify-style arrangement where some consortium of copyright holders receive some tiny revenue-share for each search that touches their content.

Or alternatively, perhaps MSFT just agrees to fork over a one-time payment as a "training use" license.

Having said all that I think NYT probably bit players here....the rumor is that OpenAI utilized libgen (pirate book site) and downloaded hundreds of thousands of ebooks as corpus data. So likely these lawsuits are just the beginning and its going to get messier before it gets clearer....