Because even generic USB devices that adhere to standard device classes use drivers? And it is perfectly possible for a device manufacturer to still have a custom driver because they want added functionality?
Ages ago I was developing the USB functionality for a device and accidentally came up with a particular firmware load which did something wrong during the initial connection of sending back & forth device identification info... on any Windows machine (98, 2000 & XP) we tested it on that you plugged it into, the device discovery would fail, so you'd unplug the device and move on... and 3 minutes later the PC would seize up (no BSOD oddly enough).
For some reason I never reported the bug, nor did spend any time trying to figure out what bits of my code were breaking Windows, I just solved my problem, made the device be recognized by Windows and move on.
When parsing any protocol or format, it is often possible for there to be unexpected cases which weren't adequately tested which make have negative side effects. This shouldn't be a surprise, I'd just be curious to know what specific change in the new update caused this.