I mean, that's how you'd train the main algorithm, right?

I'd fully expect to see these interventions fed back into the algorithm so Google can better predict "this search term is likely to be targeted by partisan or otherwise suspiciously motivated actors".