Child safety features on the most popular social media apps often don’t work as advertised, a new report found.
Researchers at New York University and Northeastern University tested dozens of safety features promoted by Instagram, Snapchat, TikTok and YouTube in recent years, as the products came under fire for enabling adolescent loneliness, bullying and sexual exploitation. The study found that in some cases, the safety tools appeared to be missing altogether, while in others they were broken, easily circumvented or difficult to find.
Snapchat, for example, allowed adults to send message requests to children they didn’t know and suggested that teenagers befriend adult strangers. Instagram, too, prompted teen accounts to connect with unknown men. And TikTok, after promising to remove content that promoted eating disorders, recommended searches to teen accounts such as “how to pretend to eat your food.”
The findings, many of which were replicated by The New York Times, come in the midst of an intense backlash against the social media industry. A raft of lawsuits claiming harms to young users could cost the tech companies billions of dollars, and several countries have announced bans on social media for children under 16.
Several tech leaders are set to testify to Congress next month about their apps’ effects on children. The companies say they have made their products safer over the years, pointing to parental controls and time limits for young users.
The new research highlights the gulf between the companies’ assurances about child safety and the experience of parents and teens online. The researchers combed through hundreds of company statements about the features and found that the language often implied that the platforms were doing much more to prevent harm than they actually were, said Lexie Matsumoto, a graduate student in computer science at N.Y.U. and an author of the study.
“They have language in there like, ‘We make it difficult’ or ‘We have measures to protect against this,’ without ever really describing those measures,” she said.
In 2023, Snapchat said teens needed to have “several mutual friends in common with another user” before they could show up in search results or be suggested as a friend for that person. And in 2025, the company announced more “protections for younger users that help prevent strangers from being able to find their profiles and connect with them.”
But when the researchers signed in to adult accounts on Snapchat, they could find and send friend requests to teen accounts by searching for the teen’s username. The app also frequently recommended unknown people to teen accounts. The Times replicated these findings nearly a month after the study’s authors reported them to the company.
A Snap spokeswoman said that the researchers, who were intentionally trying to bypass safeguards, did not represent typical users. She added that the app warns teen users to be cautious when someone outside their network tries to contact them.
Got a confidential news tip? The New York Times would like to hear from readers who want to share messages and materials with our journalists.
On Instagram, teen accounts are private by default, a feature the researchers praised. Still, when testing a new account for a teen girl, they found that a page of recommended people to follow listed almost entirely profiles of what appeared to be adult men. A new teen account created by The Times also saw suggested profiles of unknown adults, both men and women.
A Meta spokeswoman rejected the idea that Instagram’s safety tools were broken. “The reality is that with Teen Accounts, teens are seeing less sensitive content, experiencing less unwanted contact, and spending less time on Instagram at night,” she said.
Three platforms — YouTube, Instagram and TikTok — created notifications meant to limit the time teens spend on the products. But the researchers found that these alerts were easily ignored. When a teen YouTube account reached a 60-minute limit for viewing short videos, for example, the platform immediately offered a link to change the limit, as well as an option to “ignore limit for today.”
A YouTube spokeswoman said that parents could set up time limits for short videos that their children cannot bypass.
All of the platforms promised to reduce teens’ ability to find content related to self harm or eating disorders. But on TikTok, Snapchat and Instagram, users could get around this by slightly adjusting their search terms — such as using “eating disorser” instead of “eating disorder.”
On TikTok, a teen account that had looked for eating disorder content would then be prompted to search for terms like “mentally suffering,” “razer blade skin” and “anna food tips,” a common shorthand for anorexia, the researchers found.
When a Times reporter tested the platform with a teen account in late June, after the researchers had informed TikTok about the problem, the company had partly fixed it: The recommendations for harmful search terms were gone. Still, after a few searches about dieting, the page for suggested content was inundated with images of young women showing off their clavicles, congratulating themselves for not eating and offering “toxic wl motivation,” referring to weight loss.
A TikTok spokeswoman said the app has more than 50 safety features for teens turned on by default, and the company regularly removes content that glorifies disordered eating. TikTok has additional restrictions for users under 13, a feature that the researchers found worked well.
Anneke Buffone, a psychologist who worked at Meta until January, when she left to found a youth-safety nonprofit, said that social media platforms didn’t invest enough in trust and safety teams, leaving those workers stretched thin and unable to fully fix problems with these features.
