Persuasive ad campaigns are an attack on individual choice. The goal is to convince someone to do something: to buy, to vote, to labor, to behave, etc.
If the attack results in an individual choosing something that benefits them, it may not seem evil. However, the difficulty arises in how one gauges benefit. Can advertising bodies be trusted to or even capable of measuring the outcomes individuals' of choices? Can governments?
I will assert that only individuals can be trusted or even capable of measuring choice outcomes. (Democracy is based on this assertion, otherwise voter's choice wouldn't matter). If this assertions holds, any attack on the individual's capability of gauging choice is an attack on democracy and their liberty.
Democracy is designed to be resilient. Essentially the democratic process is a boosting method for classifying the best choices for a population. [1] (Our forefathers weren't dumb). The citizen's choices just have to be slightly better than random to collectively form a single more reliable choice. If citizens' collective choice accuracy falls below 50% the process begins to fail.
Therefore any advertisement that subverts individual, rational choice is evil.
How would one determine if an advert is subverting rational choice? I don't know, but I wouldn't mind them being purged with impunity.
[1]: https://en.wikipedia.org/wiki/Boosting_(machine_learning)