As a theorist, I often motivate the need for strong cryptography via simplistic methods, such as "If this did not exist, your online bank transactions would be vulnerable". This is of course true if "this did not exist" means that we are in Minicrypt (or some weaker "world"), but I also sometimes use this to motivate switching off of particular weak cryptographic primitives. In this setting, it seems like social/legal conventions would mitigate the potential harm to the user --- for example, simple ways to attempt to leverage the total break of a cryptographic primitive (say transferring a ton of money into your personal bank account) would easily be undone through legal avenues.
In truth, I suspect that most of the hypotheticals that I can think of for how to nefariously exploit a broken cryptographic primitive in the wild would be easily reversed by law enforcement (perhaps I am just a bad cybercriminal though). That, coupled with my perception that "most" cybercrime does not rely on breaking the "actual" cryptographic primitives that are used (and rather implementation errors/social engineering), makes me curious if cybercriminals have ever attacked underlying cryptographic primitives as part of their criminal enterprise. I would be most interested in novel attacks / attacks against primitives that were thought to be strong beforehand, but as I suspect these do not exist I would also be interested in attacks against primitives that are known to be weak, provided they were against some notable company (where "notable" essentially means "something an average person may have heard of").
I want to exclude attacks by governments in this question, as these (especially via the NSA) tend to be fairly well-documented already, and seem to mostly sidestep the "law enforcement could/would undo the damage" argument.