Social media companies claim their platforms are safe for kids, promoting safety features that include restricting searches for harmful content, limiting messaging and warning of out-of-network contacts.
TikTok, for instance, is supposed to block minors from searching for material about disordered eating and self-harm.
But research from Northeastern University found that instead of blocking such harmful material from appearing on an account registered to a minor, TikTok began suggesting it.
Researchers found that 35 of 86 child safety features tested across major social media platforms worked as promised, while 51 failed to meet their standards.
“It recommended that our teen account look up ‘anna food tips’ [“anna” referring to anorexia] and ‘how to pretend to eat your food,’ both drawn from pro-anorexia communities, alongside ‘mentally suffering,’ ‘losing yourself to mental health’ and, most disturbing of all, ‘razor blade skin,’” research led by Laura Edelson, assistant professor of computer sciences at Northeastern, reported. “These were the product’s own recommendations, served to a child, not phrases we went looking for.”
The safety features that social media companies have enacted are meant to address a range of risks. Other researchers have used “the 4Cs” to evaluate online harms to children. These Cs are content (exposure to age-inappropriate material), contract (commercial, monetary or data exploitation of a minor), contact (harmful commenting or direct messaging) and conduct (harmful behavior by a user).
In their study, Edelson and colleagues added two additional categories in evaluating safety features. One of them, “Circulation,” addresses the risk of an underage user’s content circulating beyond the intended audience. For example, Edelson said a teen sharing pictures of her gymnastics team at an event is very different from an adult male who has no connection to the teens sharing such pictures. “Compulsivity,” meanwhile, addresses designs like autoplay that make social media difficult to put down.
Researchers tested 86 safety features, including search content filters and messaging blocks, for children that were promoted by Instagram, TikTok, Snapchat and YouTube. They chose those four platforms because of their popularity among youth, Edelson said. Researchers set up fake teen and adult accounts on each of those platforms to evaluate the safety features.
Each safety feature was tested on two criteria: whether it functioned as claimed by the social media companies, and whether the tool was ever shown to the child user in scrolling. Successful safety features met both criteria, according to the research.
The safety features also needed to be on by default or easy to activate, be resilient to normal teenager use and demonstrably function, researchers said.
Only 35 of the 86 safety features were successful, according to the research and the remaining 51 safety features failed in several ways.
Researchers found nine safety features were “missing” or could not be triggered at all, even after following the steps described by the social media company. Edelson noted that Instagram has a “prompt-to-reconsider” feature where the platform will ask you whether you really should type, as Edelson said, “some terrible thing” such as profanity or harassment as a comment. The feature was not triggered when a teen account used explicit language and clearly bullied another teen, according to the research.
Thirty-four safety features were “broken,” meaning the tool existed but either failed in a way that defeated its purpose or was what Edelson called “trivially circumventable.” Of these thirty-four features, 12 safety features were both “broken and buried,” according to the research, meaning that they were both nonfunctional and “effectively unreachable” by either being located under layers of settings or were too difficult for the average child to configure.
The remaining eight of the 51 safety features were simply “buried.”
Although each platform differs, they share some commonalities.
All of the “conduct” safeguards meant to regulate how users treat one another and prevent or detect cyberbullying failed on all four platforms, researchers noted. Only 1 in 3 tools to curb compulsive use, such as screen time limits and reminders to take breaks, were successful, according to the research.
Many safety features relied on a list of blocked terms, but simply misspelling one of those words could effectively bypass the feature.
For all the failures, there were some successes.
Instagram automatically sets teen accounts to private when they are created, for instance, while TikTok for Younger Users, those under 13, is view-only, and its users cannot comment or message other accounts.
Northeastern Global News, in your inbox.
Sign up for NGN’s daily newsletter for news, discovery and analysis from around the world.
“The number one lesson that I take away from this is that making the default experience the safest one — basically making safety the default — is something that works very well,” Edelson said. “Something else that works very well is, for the youngest users, instead of leaving the risky service in place and then trying to build a fence or a guardrail around it, just removing that risky service.”
Northeastern Global News reached out to the companies that run the four social media platforms for comment on the study. YouTube and TikTok responded.
“We’ve spent over a decade building industry-leading parental controls, which is why 84% of parents who have used YouTube supervised account tools said they agree that these tools give them confidence that their child is accessing a safer and more controlled digital environment,” a YouTube spokesperson said. “We will continue to strengthen these protections and innovate to protect families who use YouTube.”
Added a TikTok U.S. spokesperson: “Teen accounts on TikTok come with over 50 preset safety features and settings automatically turned on, with additional choices for parents through our easy-to-use Family Pairing tool. Our internal review confirms these features are working as intended, and we welcome the opportunity to help the authors of this report better understand how our app works.”
NGN reporter Hannah Morse contributed to this story.
————————————————
