In the standards for the types of RCD used in Europe, there are three categories:
- Type AC. This only responds to AC leakage currents, of the sort likely to arise in appliances and wiring that have no electronics. It can be purely electromechanical in its internal design (a simple transformer captures the difference between currents flowing in the two wires of the circuit and a third winding on the transformer triggers the mechanism). Most installed RCDs are of this type.
- Type A. This responds to AC leakage currents, and also to pulsed DC currents - as can easily be caused by modern appliances with electronic drive to motors etc. They can be built using the same sort of transformer as before, just needing a bit of electronic filtering to give the required behaviour, so don't necessarily cost any more. There is a slow move to using this type for most applications.
- Type B. This responds to AC and pulsed DC like the Type A, but also to pure DC leakage. Since pure DC doesn't work with transformers, a different mechanism is needed to detect the DC and so Type B units are fundamentally more expensive to make (some designs are also physically bigger).
One of the issues is that with pure DC leakage not only will Type-AC breakers upstream fail to trip as a result the DC itself (which might be a minor and unimportant fault), but they will also be 'blinded' (the magnetic core of the transformer becomes saturated) and they won't then trip on any other kind of fault elsewhere in the installation. However, the amount of DC required to make this happen in practice is usually quite a lot more than 6mA. Old test instruments for measuring the earth fault loop impedance used to do this deliberately to avoid tripping RCDs during the test - but they needed to use quite large currents, and many modern RCDs will in fact trip when one of those "no trip" test instruments is used (modern ones do 'no trip' in a different way).
I don't know how the USA standards for GFCIs compare (GFCI and RCD are fundamentally two words for the same thing, but detailed specifications differ).
For EV charging, since specific regulations were introduced (published in 2008 for the UK regulations, though the text comes from Europe-wide CENELEC standards) there has been a requirement to use at least Type-A; Type-B is obviously better, but the regulation didn't require it except if it was known that the installation was specifically subject to DC leakage greater than 6mA. The RCD can be integral to the chargepoint or external, but it has to comply with relevant standards (EN 61008-1 etc), so although many chargepoints like OpenEVSE have an RCD function, they usually don't meet all those requirements and so a standard RCD has to be installed on the circuit in addition. Most of these built-in RCD functions simply open the EVSE's existing contactor when they 'trip', so they work well on small leakage faults but those contactors are not rated to disconnect a high-current line-to-earth fault. Some models of chargepoint commonly sold over here have a standard RCD (such as you might have in your Consumer Unit (= breaker box)) incorporated in the chargepoint housing and so don't need an external one.
In the new regulations recently published (and taking effect January 2019) this has been turned around: now Type-B is REQUIRED unless the chargepoint ensures that any DC leakage is less than 6mA.
In the UK at least, Type-B RCDs are hard to find, very expensive, and won't easily fit into existing enclosures. The obvious solution is to implement the DC detection in the chargepoint - this can be simple and lightweight as it's not trying to do all the work - and then we can continue using economical Type-A RCDs to protect the circuit. However, there's not much sign as yet that the chargepoint manufacturers have this in hand - maybe I will be surprised and there will be a flurry of new product announcements in the next month, but if not there will be big problems with new EVSE installations come january.
Aside from the legal position, the question is whether pure DC leakage is really more likely to occur with EVs than with any other modern equipment that has electronics in it. One argument I have seen is that if there is a short circuit between pilot and neutral in the charging cable, then the EVSE will drive about 12mA around the neutral-earth loop. That certainly is a pure DC leakage greater than 6mA, but whether it is enough to actually cause trouble is debatable.