I'm pretty sure US startups also have to respect copyright law when mining data on the internet?

The U.S. has Fair Use, which adapts to novel cases. The EU has enumerated exceptions to copyright.

As a simple example ripping a CD is tortuous in the UK. It was briefly allowed for format shifting, but that was deemed too antithetical to capitalism, or some shit.

Schools have to have PRS (songwriters) and PPL (musicians) collection agency licenses if they have TV because the license they pay for TV doesn't cover the music "performamce" eg of an adverts soundtrack.

Individual solo workers can't listen to a radio at work unless they're unable to be accessed by the public.

You can time-shift a TV show (say), but you're only allowed to watch it alone, and you can't watch it twice.

UK is draconian in the extreme.

Personally I'd favour a 15 year term, and only for works that have an identical un-protected copy lodged with a central agency (at a high cost), or are DRM free. Transfer/download rights for copies bought on streaming services; format shifting, relaxed time-shifting, more educational use rights, etc..

Of course they do, but US startups are facing far fewer restrictions on mining data published in their home market.

Large internet companies like Google and Facebook will face even fewer restrictions because they have the clout to force economically viable terms on publishers.