Many principles of "ethics" are designed to protect autonomy. Regulations concerning human subjects in research take a broad view of what counts as "coercion", to the point of including excessive rewards that make it difficult for potential subjects to refuse. Programs to reduce excessive birth rates were criticized as violations of autonomy, even if they consisted of advertising and TV soap operas (as I discussed in my book "Judgment misguided"). Now, some of the opposition to "nudges" is that they, too, violate autonomy, because they influence people, which is, of course, their purpose.
This concern has always struck me as a little strange, because it seemed to me that well-intended attempts at influence were a drop in the bucket of all the things that influence decisions, including much more blatant and less justifiable attempts by others. Nudges from government are minor influences compared to marketing strategies by corporations, or expectations of families. Philosophically, one might even ask what is left if all these influence are removed. We are all influenced by our culture.
But the drop-in-the-bucket argument is weak at best. If autonomy violations are wrong, then the existence of large amounts of wrongness in the world is no justification for adding more.
A more interesting property of ethical concerns about autonomy violations is that all (or almost all) of the alleged violations are actions. The operating moral principle seems to be that we should not take action to influence the decisions of others, but we need take no action to prevent autonomy violations that already exist. If women in poor countries are being pressured by their husbands and in-laws to have more children than they want, those concerned about excessive undesired births are under no obligation to try to prevent such pressure. Similarly, if advertisers induce people to eat lots of unhealthy food, the operating principle does not oblige is to try to stop them, although it prevents us from trying too hard to nudge them in the other direction.
Thus, the problem with all the existing influences is not that they are a large bucket. Rather, it is that the concern with autonomy arises from a moral principle that distinguishes acts and omissions and is thus not primarily concerned with consequences. That is, it is a deontological rule. If people follow this rule, as compared to one that brings about the best consequences, then it makes things worse.
The stereotype of the Jewish mother is someone who does not hesitate to butt into her children's affairs, or those of other people, for their own good. Perhaps we need more Jewish mothers. They won't always be correct about what is good for other people, but trying to act in this way might have a higher success rate than doing nothing in the name of autonomy, and thus ceding control to those who base their efforts to influence others on their own self-interest or narrow ideology.