Hmm. If you trust users to discern between malicious and non-malicious emails (that can contain things like "the mandatory new tool to fight malware" as you say), I think that implies you trust users to discern between malicious ads and non-malicious ads.

And I don't see any reason that the spam/malicious-content filtering tools used to filter email would not be used to filter ads (with appropriate modifications of course).

So this leaves me unconvinced that Zoom web ads are any more susceptible to being malware vectors as compared to ads anywhere else (or other tools that host user-generated/third party text, like email, messaging, and social media).