It's possible that people looking back will consider that the mistake was putting all the content online. Perhaps even upstream of that: the first mistake was digitizing things. The music industry certainly didn't realize when they adopted CDs that they were starting down the path to self destruction... the newspaper industry likewise didn't notice how profound taking their newsprint product and packaging it as HTML would be...

And now we're unleashing ML training on all that digital, online data. Which industries will discover that this is the thing that means putting your data online, digitally, was a mistake? Certainly artists are feeling it now... maybe programmers, too, a little.

So how do you put the genie back in the bottle? Live performances, with recording devices banned? Distribute written material only on physically printed media - but how to prevent scanning? Or just escalate the DRM war - material is available online, but only through proprietary apps on locked down platforms?

Or is this going to take regulation - new laws to protect copyrights in the face of ML training?

It wasn't always the case, that you could assume that if some information exists, it should show up in a single search. That's an expectation we invented only about 25 years ago. It's possible that the result of all this is that we figure out that we can't actually sustain the free sharing of information that makes that possible.

The problem is, to borrow a phrase: information wants to be free...