Rampant fraud = lack of trust in your product/service, which can destroy your whole business. The internet fraud you hear about is only the tip of the iceberg, and there's much more of it that is prevented by various companies (either Google, Cloudflare, etc.)
They dominate search traffic (in the markets/countries they exist for). Businesses don't use AdWords because they "trust" Google. They suffer through using AdWords because it's the only game in town.
"If advertisers believed the company's operation were fraud-filled, they could take their money elsewhere and the business would falter."
That's it, in a nutshell.
(and if they had 100 people working on this in 2015; you better believe that number is much higher now)
Maybe Google had more competition in 2015.
Although I find even that difficult to take seriously.
Where else would they "take their money"? Facebook? I doubt that Facebook ads show up in web searches.
And where would that be, exactly?
It's in in Dutch, but it's the typical misleading, copyright infringing ad you would find on shady porn sites.
Why does Google allow this ad? Don't they have the competence to filter it? It's the same company creating a great spam filter on gmail and leading in a lot of AI fields. Surely they could. But they just don't care when they get paid for it.
On the other hand, when they don't get paid or it negatively affects them, they're always happy to make changes "to protect the users" (e.g. changing chrome so adblock plus stops working etc.)
The most simple explanation is because they make money with it, and get away with it. A good question is why they get away with it.
As for people in The Netherlands, my mother got scammed by a fake locksmith. He asked her to pay with PIN, and he did not fix the problem (which we found out eventually). Total scumbag, yet he was first hit on Google. Here's some good advice on how to find a good locksmith (slotenmaker) in Dutch [1].
[1] https://www.consumentenbond.nl/inboedelverzekering/slotenmak...
Think about it would take to detect it algorithmically. Even if an image is copyrighted there is no mechanism to tell if the source actually has permission from it especially with the "born copyrighted" doctrine. Even if there was some vast self defeating registered copyrighted images database (containing images of everything copyrighted) it woukd be possible to distort images to be human recognizible but not machine recognized.
The "alternatives" of demanded manual ad control would mean massively disadvantaging smaller business by marginal costs. Oops.