Proving bad faith or that someone "knowingly materially misrepresents... that material or activity is infringing" when it's completely automated sounds very difficult.
One has to wonder if that's part of the intent.
"Oh, haha, sorry about that. Our automated tool screwed the pooch on that one. We promise we'll do better next time."
My limited understanding is that these companies just use google search apis to try to find search results matching keywords. Then they file requests for every matching url. This is how obviously wrong requests show up.
There are also cases where requests appear to be malicious, but there are really no consequences since you (iirc) have to prove bad faith which is next to impossible and since everything is being adjudicated via third parties, there is really no incentive.