No matter how old you are YouTube’s AI age verification system should concern you

Earlier today, YouTube began rolling out an AI-powered age estimation model, which they say is part of enhanced protective measures for children under 18. If an account is flagged as potentially belonging to someone under 18, YouTube will add “age-appropriate” protections to it, such as the inability to watch age-restricted content, showing only non-personalized ads, activating “digital well-being tools” such as break and bedtime reminders, and providing privacy prompts when uploading or commenting on videos. And if an account is labeled incorrectly as belonging to a teen, a user must upload a selfie or a picture of a state-issued ID (like a driver’s license) or a credit card to continue using the service.

All this sounds great on a surface level, but privacy experts are sounding the alarm. A Change.org petition with over 75,000 signatures protesting the change warns that this could pave the way for “mass surveillance and data control.” And while that may sound a little hyperbolic, I actually don’t think the petition writers’ logic is actually all that extreme.

Privacy concerns

Who do you want having your data?

The most immediate concern I think most reasonable people have with this so-called feature rollout is how safe their private information is if they are selected as an account that needs ID verification. While I think most would agree that an uploaded selfie probably doesn’t expose most people to any glaring data privacy issues, if Google decides that they need a copy of your driver’s license or a credit card, that could pose a huge risk of identity theft.

In a world where hundreds of driver’s license photos were leaked only hours after a controversial networking app hit the App Store (per the BBC), this is a very valid concern to have, and I know I would have plenty of reservations about handing my address, driver’s license number, or credit card over to Google just so I could continue using YouTube.

Should AI decide who you are?

I don’t think so

Pocket-lint / Glenn Carstens-Peters / unsplash

The way this technology works also poses some serious ethical questions. According to Google, the AI will scan for patterns among the “types of videos a user is searching for, the categories of videos they have watched, or the longevity of the account” to try and determine a user’s age.

And while this may be useful for accounts that only watch Miss Rachel or videos of children’s books that are read aloud, what about content like video game playthroughs or theories about the next season of anime series Solo Leveling? These videos are all in my viewing history, but they would also look right at home in a 16-year old’s viewing history as well. Why should who I am, or who a teen is, be defined by our viewing habits? And perhaps more importantly, why is Google letting an AI decide whether a specific account’s viewing history aligns with a stereotypical demographic?

I have a teenage family member who watches videos almost exclusively about botany, gardening, and classical music, because that is what her interests are. Somehow, I doubt that YouTube will be flagging her account as underage (even though she is), specifically because her viewing history doesn’t fit the stereotype of what a teen watches. However, because mine probably does tick a few boxes, I’m more concerned about my own account needing verification than hers, if we’re going solely off what an AI thinks a teen might be interested in. People are complex, and it simply rubs me the wrong way that an AI is making assumptions about who someone is based solely on their YouTube viewing history.

What will happen with the data about your identity

I don’t think it’s just going away

Firmbee / Pexels / Pocket-lint

It’s fair to say that the almighty algorithms behind YouTube, TikTok, Spotify, and the litany of other personalized services I use on a daily basis know a fair bit about me, and can probably guess that I’m interested in fantasy novels, anime, video games, and technology. However, now that we have confirmation that AI is sweeping things like our YouTube search history to look for identifying patterns to try and guess who we are, I can’t help but be very concerned as to what else could possibly be done with this information.

Data collection for the purposes of advertising is nothing new, but the idea of a powerful AI creating profiles for users based on trends or stereotypes troubles me quite a bit, as I can imagine it being used in very nefarious ways in the future, not only by potential data thieves, but by the tech companies implementing these changes themselves, which seem to have no issues crossing ethical boundaries in the name of increased profits.

Moving even further into a dystopian nightmare, if the government wants to subpoena your digital AI records, in theory they could potentially gain access to your information as well, which would be extra easy for them if you had to upload an ID to use YouTube when you were a senior in high school.

Pocket-lint / YouTube

Though the situation does seem a bit hopeless, users are pushing back, threatening to cancel or delete accounts if they are asked to provide identity verification, at least according to Dexerto. Unfortunately, we’ll have to wait and see if these protests will have any effect or if YouTube’s almighty AI-powered account scanning is here to stay.