We withdrew the ads after YouTube’s approval and before they could be published, to ensure that they did not run on the platform and were not seen by any person using the site.
Eh, I want moderation to be a human-driven effort. How would you expect it to catch content without users flagging it??
Oh, AI? No thanks. That's how you end up with false positives. That's how you end up with Israeli war crimes getting flagged and removed.
Sorry, but you didn't test anything if you didn't let users see and flag the content in question.