I find this part a bit surprising. Brand advertising surely cares about fraud when it’s done on a cost per impression model. But I expect that brand advertising’s effectiveness is more related to placements, not impressions. For example, Nike once famously reduced the number of swooshes they used because they thought that having too many diluted the brand. Similarly, a brand placement somewhere like a movie or an article might have high impact despite only being seen once per viewer, whereas a placement on a site that gets lots of intra-site clicks (like Amazon or some mediocre review site) will get many more impressions per viewer but likely less impact per viewer despite a higher number of impressions.
So, with a cost per placement model, I would expect the simple forms of fraud to be much less relevant.