[andrewla]

The article talks about the possibilities of malicious cloning of these tokens by third parties, but fails to identify the much more common use case, and one that makes this scheme useless for age verification.

It's one thing to be concerned about someone stealing my credential, but another to prevent the transfer of these credentials, especially if they are limited use credentials.

The entire point of age verification systems is to prevent minors from accessing certain resources. I think we all know that this is basically impossible; but what these various governments and social media companies want to do is to make it high friction to do so.

The highest friction version of this is that the credential ties to a real world identity somehow; maybe locked behind legal barriers, etc., but if a minor is caught using someone's credential, then the person whose credential they are using can be investigated, and, if necessary, charged with a crime roughly equivalent to providing alcohol to a minor. Without the possibility of real world enforcement, none of these identity solutions can possibly work.

Keep dreaming of a technological solution -- there is none that does not lead to the world that FIRE is warning about, except to accept that we can only make a solution "good enough" and leave it at that, without expanding into full on identity verification. The solution here is likely to just try to provide better abilities for parents to monitor and limit their children's use of the internet. Let individual parents decide on the level of harm that they are willing to accept, and accept that there will be ways to work around this even if parents are vigilant, but just try to reduce it on the margins.

[Aurornis]

Yes, this is the part of the issue that is so frequently ignored: Anonymous age verification schemes are easily defeated through proxying because there wouldn't be any consequences for selling your tokens. "Install this app on your phone and we'll pay you $1 per day" and it will mint your anonymous identity tokens and send them off to kids who want to buy them. If there's no way to track the tokens, there is no possibility of negative consequences.

So the schemes always start introducing features to reduce the anonymity of the tokens or make them more trackable in some way:

> The highest friction version of this is that the credential ties to a real world identity somehow; maybe locked behind legal barriers, etc., but if a minor is caught using someone's credential, then the person whose credential they are using can be investigated, and, if necessary, charged with a crime

Which requires that these identity tokens not be anonymous age-verification credentials. They become a traceable identity token tied to your government-issued ID.

[DennisP]

> They become a traceable identity token

Not if you use a challenge-response protocol where the client returns a zero-knowledge proof of age, where the proof incorporates a random string sent by the website.

The traceable stuff is private information that the website never sees. If a minor is caught with it, then law enforcement has local access to the minor's hardware and can probably view the private data.

At that point, the private key can be put on a public revocation list. The zero-knowledge proof can include a proof that you're not on the revocation list. Once you've been revoked, you have to go through the hassle of setting this all up again, which might be enough incentive to keep it reasonably secure.

[pastel8739]

This doesn’t stop the scheme the parent proposes, where adults install some proxy on their device and challenges are responded to on the parent device. Then the private key never leaves the parent device and all the child device has is the proxy software, which could be set up to not log any identifier of the key that it used

[7e]

Trusted computing fixes this.

[pastel8739]

How so?

[7e]

Trusted computing fixes this up to the analog hole. Which is as much as you can expect.

[hacker_homie]

I thought a solution to this would be to use a physical smartcard to store the certificate(perhaps on your government ID). if the protocol is a challenge/response and the private key never leaves the card it would make proxying without the physical card more difficult.

[wolvoleo]

Yeah great idea, having to get out your government ID every time you want to use a website.

[pastel8739]

If the smart cards required some human input to perform a signature maybe this could work. Otherwise there is nothing stopping someone from selling use of their card via some proxy software

[7e]

Trusted computing solves this problem handily.

[AnthonyMouse]

> but if a minor is caught using someone's credential, then the person whose credential they are using can be investigated, and, if necessary, charged with a crime roughly equivalent to providing alcohol to a minor. Without the possibility of real world enforcement, none of these identity solutions can possibly work.

They don't work even then.

Suppose you completely eliminate privacy on the internet and require every domestic site to collect the name and social security number of everyone who visits. Then a child uses an adult's ID, regardless of whether it's with or without their knowledge. Is the child going to inform on themselves? No. Is the adult, when they don't even know about it? No. Is the adult, when they provided it on purpose? No.

That constitutes the entire set of people who would typically know that the person using the device isn't the person on the ID.

On top of that, we can punch an even bigger hole in it. Search engines, among other things, index other sites. Google is obviously the biggest but there are many others -- Bing, Marginalia, Brave, Swisscows, Yandex, Perplexity, Baidu, etc. They're run by adults and most of their users are adults, who reasonably expect to be able to turn off "safe search" if they want to. So some adult at each search engine would have to provide their ID to the crawler so it can index things inappropriate for children and show them to adult users. It would therefore be a fairly unremarkable and recurring thing to see the same ID make a zillion gigatons of requests.

But then you can't use "why is this person downloading 100 things from 100 computers at once" as an indication of anything nefarious happening, and anyone can still set up a service hosted on a foreign server that will serve adult content to anyone without an ID by serving it out of a cache. (And in the case where you're invading everyone's privacy, that service would also be very popular with adults.)

[ajsnigrutin]

> The highest friction version of this is that the credential ties to a real world identity somehow; maybe locked behind legal barriers, etc., but if a minor is caught using someone's credential, then the person whose credential they are using can be investigated, and, if necessary, charged with a crime roughly equivalent to providing alcohol to a minor. Without the possibility of real world enforcement, none of these identity solutions can possibly work.

Buying alcohol for a minor implies knowledge and intent.

Getting the tokens out of a phone doesn't require the user to do any of that, the user just has to be frugal and keep the phone longer than it's supported by the manufacturer, until some local exploit is found again, and that token will be extracted and available online for everyone to use.

Parents buy those phones, phones could easily have a "user is a minor" setting (and a flag sent to all the sites that want one) with a password for parents to unlock stuff if needed. This would be set during the phones first set up, and it's done. But nope, the plan is for everyone to install a form if a digital ID on their phones, and once it's there, requiring full-name identification when registering is just one step away.

[what]

Why can’t you just sell single use codes at gas stations/liquor stores/etc and they just check your ID before sale? Of course shady places can still sell them without ID check, but we have this problem already for liquor and tobacco.