Voice cloning doesn't just copy your voice. It upgrades it.
Researchers tested widely used voice cloning models and found they don't faithfully reproduce the voices fed into them. Human listeners rated the cloned outputs as more authoritative, warmer, more polished in a customer-service way, and more human-sounding than the original recordings. Those same listeners said they trusted the clones more than the sources, and were more willing to hand over sensitive personal information to them. The study also found cloning flattens differences between speakers, shrinking the variance in accent, speaking rate, and the underlying audio embedding space.
That's a problem for every use case built on the assumption that cloning is a faithful copy: finishing an unfinished recording, dubbing a film into another language, or restoring a voice lost to illness. If the output is actually a style upgrade that reads as more trustworthy than the real thing, systems built around consent and identity verification are working from the wrong model of what the technology does. A process that smooths out accents and vocal quirks on its way to sounding "more human" is, by definition, erasing some of what made the original voice sound like a specific person.
Call it the uncanny valley in reverse: the fake doesn't creep anyone out, it just seems like the better version of you.