The USB 2.0 spec says: "A low-speed device is required to have a captive cable." However, I can't find a good explanation for why this is the case. I have seen hints towards two possible explanations:
To prevent users from accidentally using a (non-captive) low-speed cable with a full-speed or high-speed device. However, there is an error in this logic: To achieve that, the spec would only have to demand that low-speed cables must be captive. It would not have to demand that low-speed devices use a captive cable, the difference being that a low-speed device could use a non-captive full/high-speed cable.
Differences in cable capacitance. Section 18.104.22.168 says: "A low-speed device (including cable) may have a capacitance of as little as 200 pF and as much as 450 pF". Unfortunately I found it hard to find numbers for high-speed cables. Things get more complex because different cable lengths exist, but then, I don't understand why the spec disallows using these cables altogether instead of saying, if a low-speed devices has a B-receptacle then it must handle all standard cables, and only if it can't do that a captive cable is required. Also, in case standard cables have a capacitance that already exceeds 450pF (as noted, I don't know the actual numbers), then hub drivers are strong enough to handle that because they must (for fullspeed), so they could do the same for lowspeed -- and for the device, again, a captive cable could be required only if its driver is too weak for the capacitance range of standard cables.