Instagram Plans to Use AI to Catch Teens Lying About Age

[Artaxerxes II]

https://uk.finance.yahoo.com/news/insta ... 00359.html:

(Bloomberg) -- Meta Platforms Inc., under fire from parents and lawmakers over its impact on teens’ mental health, plans to use artificial intelligence to identify young users on Instagram who are lying about their age and automatically switch them into more restrictive privacy settings.

With a proprietary software tool it calls an “adult classifier,” Meta will categorize users into two age brackets — older or younger than 18 — based on the person’s own account data, according to Allison Hartnett, Meta’s director of product management for youth and social impact. The software can sift through a user’s profile, see their follower list and what content they interact with, and will even scan unsuspecting “happy birthday” posts made by friends to predict a user’s age.

Based on the software’s findings, people who are suspected to be under 18 will be automatically placed into teen accounts regardless of how old they claim to be on their profile, Hartnett said, sharing more details on the process for the first time. In September, Instagram introduced these new teen accounts, which come with more stringent default privacy settings, like limiting who a user can receive messages from and what types of content they can look at. The company didn’t disclose how accurate the adult classifier is.

The company is already moving teens into these more restrictive settings based on their self-reported birthday, but plans to utilize the adult classifier early next year to help catch people who may be trying to skirt the new rules. All users under 18 will be automatically herded into teen accounts, but those who are 16 or 17 will be allowed to change the more restrictive privacy settings on their own. Anyone younger than 16, however, needs parental consent to disable the more restrictive settings.

Properly protecting teenage users from harmful content has been a challenge for Meta for years. The company’s move toward teen accounts was applauded by outsiders who have long argued that Meta wasn’t doing enough to keep younger users safe online.

The social networking giant is facing a lawsuit from dozens of state attorneys general alleging the company knowingly hooked kids on social media and helped create a teen mental health crisis. It’s also been the target of parents who allege their teens have died from drugs or suicide due to content they’ve seen or conversations they’ve had using Meta’s products.

In 2021, Facebook employee-turned-whistleblower Frances Haugen released thousands of pages of documents to the media that showed that, among other things, the company’s own research found Instagram could have a negative mental health impact on teenage girls.

With the new restrictions in place, Meta’s challenge now is enforcing them — no small task considering kids often find work-arounds to access age-restricted material. A study by the UK’s telecommunications regulator found that a third of minors with social media accounts declare their age to be 18 or above.

On many sites across the internet, lying about one’s age is extremely simple. Steam, an online video game marketplace, simply asks users to self-report a birth date before viewing and downloading a mature-rated game. Other sites that sell alcohol, for example, just ask users to click a button verifying they’re over 21.

Meta aims to make it much harder to lie about your age on Instagram. For example, the company will flag teens who try to create a new account using the same email address but with a different birthday. The company can also check a phone’s unique device ID to guess who is behind a new profile.

See More: Instagram Rolls Out Restrictive New Privacy Settings for Teenagers

Teens who try to manually increase the age listed on their account will have to prove themselves by uploading formal identification, like a driver’s license, or by sending in a video-selfie to Yoti, a third-party service that can estimate age based on facial features. Yoti and Meta then delete the images after verification. Users could previously verify their age by having friends vouch for them, but that option has been eliminated, Meta said.

People who have been wrongly categorized as teenagers may eventually be able to appeal to the company to remove their teen account status, a spokesperson said, though that process is still under development. For now, while Meta works on improving the accuracy of its adult classifier, people wrongly predicted to be teens can manually turn them off without parental permission.

Meta’s approach to age verification is made possible by the vast troves of information people share about themselves on the app, including their social circles, hobbies and content preferences. Instagram’s teen accounts are still being rolled out in the US and several other countries, but so far, Hartnett says that teens lying about their age has been less of a problem than the company expected.

Despite Meta’s efforts, Hartnett and other executives argue that the most privacy-preserving age verification process of all would involve app stores verifying a teen’s age before they can even download certain apps. A spokesperson for Apple Inc. said that such an approach would violate the principle of data minimization, while a spokesperson for Alphabet Inc.’s Google said there isn’t a “single solution” to the problem.

But shifting the burden of age verification to Google Play or Apple’s App Store “assumes that the app store has the ground truth,” Michael Smith, a computer science professor who teaches a data privacy course at Harvard University, said in an email. That’s not solving the problem of whether or not the person has been honest from the beginning. “Aren’t Meta’s problems with age verification simply being moved to the app stores?”

[Lennon72]

Oh heaven forbid! Like any of this is ever going to work. Even IDs can be faked.

[Bookshelf]

Every time I open the news, I always find what's essentially "Here are the top 10 new ways we're going to restrict our youth!"
Every week there's some new big story about how a thing is super dangerous, and that we need to change our laws to ban youth from going near it (and strangely, these changes just so happen to serve certain interests... for example, normalizing the use of AI data collection, reading private chats, etc).