Why haven't voice clones burned down our trust systems yet?

A year ago, every ed-tech conference had someone warning that voice cloning would be the end of phone-based authentication, that scammers would impersonate your kid's principal or parent and drain bank accounts en masse. The tech got genuinely good around late 2023. We're now in 2025. The predicted epidemic... didn't materialize, at least not visibly. I'm not saying it's impossible or won't happen. But I'm trying to figure out if we just got lucky, or if the actual friction points are different than the theorists predicted. Is it that the attacks *are* happening but getting absorbed as losses we don't hear about? Or is there something about how voice authentication actually works in practice—the secondary checks, the slowness of institutions, the fact that most people don't just hand over money to voices on the phone anymore—that makes it harder to weaponize than it sounds in a security briefing? The uncomfortable possibility I'm sitting with is that we spent intellectual energy on a vividly plausible threat while the real damage vectors were quieter and already in motion. Fraudsters don't always go for the flashy con. They optimize for what works. If voice cloning doesn't have a good ROI compared to credential stuffing or other methods, they won't use it at scale, even if they can. That's not reassuring exactly—it just means the actual problem might be invisible to people looking for the dramatic failure. What are you actually seeing on the ground, either in security or just regular life?