As of now the site is already back.
The core of the problem is that there's absolutely nothing effectively preventing companies from abusing IP claims to harass whoever they want.
At least you'd expect claims to be automatically dropped when coming from an assumptive/disingenuous party. Something like "you issued 100 wrong claims so we won't listen to your 101st one, sod off". But nah.
As such, "your violating muh inrelactual properry, remove you're conrent now!!!!" has zero cost, and a thousand benefits. Of course they'd abuse it.
The role of AI in this situation is simply to provide those companies a tool to issue more and faster claims, at the expense of an already low accuracy.