On the issue of online privacy regarding recent developments on Youtube's age verification and EU's Chat Control

This article will cover two very important recent developments in the tech world. One of them concerns EU's Chat Control proposal, which has been proposed and rejected numerous times until now but somehow keeps coming back, while the other will cover the intention of Youtube to roll a form of age verification powered by AI. Both of these changes are very concerning in my opinion and should be mentioned more by mainstream media, which instead turn their spotlights in Alaska for the upcoming Trump-Putin meeting regarding Ukraine, as if anything fruitful will come out of that.

Courage firm in grievous trial, help, where innocence doth scream - Ode to Joy

We are living through a challenging period for United Europe as a concept. Yet, despite the many external dangers to Europe's integrity, more and more internal dangers seem to arise from day-to-day. One of them is the so-called Chat Control proposal. On its surface, it seems like a good idea: we need to prevent the children from the dangers of the Internet. Alright, so, how do we do that? We could tell parents to inform children about the dangerous nature of the Internet (which they probably should already do on their own). How about teaching them on school during Computer Science? (yes, I know that is already being done on some countries, but in my country that's usually 1-2 hours per year, although the CS subject in Greece is generally not thought of as an important one for the majority of primary and secondary education) Wait, what do you mean the proposed solution is to spy on all online communications?

There was of course no way of knowing whether you were being watched at any given momen - 1984

While we can all agree that we need to protect the children from the Internet, what they have been trying to implement for the past 2-3 years is in no way suitable for that cause. The regulation essentially proposes spying on all online communications, which is not online morally incorrect most people would say, but also impossible to actually enforce. In case the perpetrator lives within the European Union, it would be relatively easy to arrest them. But the Internet is worldwide, which means that even if Europe could cooperate with the local police to arrest them, it would require the surveillance of not just all European communication, thus preparing the ground for similar legislations in other countries and making everyday online surveillance the norm in the entire world in the long term.

Is it even feasible?

According to PCMag, within a minute Instagram users send ~700k reels via DMs. There is no possible way that many messages can be handled by humans: automation of some sort will have to be implemented. Let's also assume that users send ~50% that volume of personal texts during that same time period, so ~350k. Assuming that the majority (99,9%) of them aren't what Chat Control wants to prevent children from, and with a 0.01% false positive rate, that would amount to 35 false positives per minute. That's just Instagram DMs: there are dozens of other platforms out there that have to be covered, as well as SMS messages and phone calls, all of them will have to be limited too, and false positives will have to be dealt with. And that's just text: add in the equation video and images and the numbers just don't sum up.

Age-estimation just won't work

Youtube is planning to roll out an "age-verification" system for the Americas on August 13. According to them, the way this will work is that viewers will be categorized on adults and non-adults depending on the kind of content they watch. This problem fall within the field of machine learning classification and there is a huge issue with the plan as proposed: how do we determine who's an adult?

Single account, multiple users

First things first, let's get the obvious out of the way: viewers that watch channels such as Cocomelon will probably be assumed to be a kid. But what if that is a shared account between a parent and their child? (people that young shouldn't even be allowed by their parents to have access to the Internet, but that's just my opinion). Does it make sense to have to use age verification just because your kid watches channels targeted at children? And if you verify that you are an adult, doesn't that defeat the whole premise of the system, since your kid will now be able to watch Youtube unrestricted?

Have you heard of Youtube Kids?

All the way back in 2015, Youtube launched Youtube Kids, which is intended to be a family-friendly version of Youtube. However, there are still quite a lot of children-targeted channels that operate on Youtube and not Youtube Kids. Apart from that, many parents don't even create a separate Kids account for their children, so why should regular Youtube viewers have to pay the price for those few parents that in the end don't care to protect their children?

Artificial borders

I mentioned above some content that is targeted to children-exclusive audiences. However, there's also content that isn't targeted against a certain audience and thus can't be used to determine whether the viewer is an adult or not. Take for example video games like Minecraft, that both children and adults watch. There are 3 possible ways to handle this kind of content:

Mark the content as children-related, bringing up false-positives
Mark the content as adult-related, bringing up false-negatives
Don't use that content to determine whether the viewer is an adult or not, reducing the training data and thus overall accuracy

Why that much hustle mid-summer season?

I don't understand why everyone, the U.K., the E.U., Youtube, are trying to pass all these major changes mid-summer, when everyone is in vacation and there is nobody to oppose such actions. Oh wait...