My opinion: A big problem that could compromise the stability of the game. (Cheaters)

This doesn't work in the USA.

The government does not provide any form of national ID card or number. (There's a lot of politics there.) The closest equivalent, Social Security Numbers, are not available to all residents. Even then, the government will only verify a person's SSN if you register them as an employee and report wages for them, so there's nothing to stop a botter from randomly entering MY SSN instead of their own. There are some background check agencies and the like, of course, but such verifications quickly become expensive.

At the same time, there is no single government ID card you can expect any US resident to have. If you want me to upload a scan of my Drivers License, what if I don't drive? My passport? What if I've never needed to travel internationally and haven't applied for one? We have a lot of different types of ID, but none are remotely universal.

And all of this avoids the fact that most americans will find these measures extremely heavy-handed, and be averse to them. We're not conditioned the same way that certain other populations are, relative to providing government identification for random online services, and we're actively conditioned to consider our SSN to be 'identity theft' level information (because it is). Nobody but banks, creditors and your employer should ever get that number.

The best you get is phone numbers, with text message verification, such as twitch and twitter use. This would offend far less people, and that... honestly is almost practical, but for various reasons, it's not exactly a silver bullet either.

EDIT: I realized that I haven't covered facial recognition. There are serious problems with this use of facial recognition related to racial biases in the technology's training sets and designs that meaningfully affect even the police ID databases. Facial recognition is meaningfully less accurate for subjects of certain non-white races. Aside from being invasive and off-putting, you don't want the reputation (or support load) you'd start to get if this sort of thing went wrong. Even without that factor, the support load required to make this technology work in the 10% case won't be cheap, nor will getting a version of it that's hard to game.

To add to all this, there are some existing discussions of this topic that have more complete discussions about how anti-botting can or should work. They go into identity verification, automated detection, GM presence, user reports, and other related concepts. Here's one.

I think smth like Nvidia's Shadowplay feature (but built into the game itself) getting triggered once a player hits "report" would streamline the process for GM and would also prevent false reports, like those that plagued NW. And you could then even penalize the false reporters.

But all the face/identity tracking seems a bit too dystopian for me. I'd hope we can push that kind of stuff away for as long as possible. Especially considering that some places have already implemented that shit.