If I remember rightly, what Prolific did was worse in some ways because it rendered genuine but older-revision chips that they'd actually made non-functional with current drivers. (Also, some of the CH340 clones are genuinely really badly done and can't understand register writes that differ even slightly from what a particular version of the official driver sent.)
I found both cp2102 and ch340 seem to use less power too than ft232r.
At least there are some boards I can power with the 3.3V output of the cp2102 and ch340 dongles, but which will brown-out with the old ft232r ones.
It's my go-to part for this purpose also (and I don't believe I've encountered any clones so far, knock on wood).
Another part I've used in a high-volume application is the Holtek HT42B534 which is great because it's CDC class and hence doesn't need a driver for Win/Mac/Linux. It's EOL sadly. There's HT42B564 which is a HID-class replacement. The other alternative is using a cheap micro with USB interface.
Aha, I didn't know about the CP2102. Looks like I'll have to give one a whirl, and there's even a breakout board at Adafruit. I've used relatively few VCP adapters since I've been using microcontrollers with built-in VCP.
I was nervous about the CH340 because the Windows drivers seemed to come from some weird place in China, but maybe US sourced drivers aren't any more of a comfort in these times.
If someone wants to quote me for a full reverse engineering of those drivers then I'd be interested in (crowd)funding it. But in terms of risks to worry about, it's pretty low down on the list IMO.