Most people in the world never faced any serious threat of jail time for copyright infringement. The absurd punishments handed down to average users is purely an American thing.
The contrast against the weirdly punitive American justice system is not the problem with AI companies.
The reality of the situation is that copyright is and always has been a horseshit system for its purpose. The entire concept of "property" is one that applies to physical goods, that are in limited supply. It does not apply to things that are abundant and ubiquitous, i.e. no one owns the air because the air is wildly abundant and everywhere.
Information: music, art, storytelling, is not physical matter. Unlike physical matter, a story can be instantly copied by as many people as can hear it, because information does not obey the same scaling laws as physical matter. Vinyls and recording equipment started exposing how infinitely copiable information is, and computers and the internet really drove that point home.
Copyright though, has always been a dumb fucking system that hamfistedly tries to apply the ownership laws of physical properties to that of information. It does so by forcing artificial scarcity on that information and creating all of these walls and systems to maintain that scarcity.
Copyright has never been fit for purpose, and the better we get at processing information, the more evident that becomes.
AI companies are problematic because they are literally burning huge amounts of resources to replace humans while there are no mechanisms in most societies to ensure humans still get resources if the robots are better than them. That's the actual problem with AI companies, and it's fundamentally a problem with capitalism.
It's far faster. Ripgrep has to search every file exhaustively at query time. Windows Search indexes every file at write time (or a background job) so the search results are near instantaneous ... at least, that's how it used to work. I don't know what happened to it over the past 5-10 years.