In 2011, danah boyd, then a researcher at Microsoft, reported that millions of children under 13 were on Facebook, despite the service’s minimum age requirement of 13. She also found that many parents knew their children were using Facebook and that some actively helped them misrepresent their ages.
Fifteen years later, age restrictions are still being widely circumvented. According to a Common Sense Media survey released in October 2026, half of American children ages 9 to 12 used social media on a given day, spending a median of an hour and 25 minutes on it. Among platforms used at least monthly by tweens were TikTok (32%), Snapchat (16%), Facebook (15%) and Instagram (13%).
In 1998, Congress passed the Children’s Online Privacy Protection Act (COPPA), which requires covered online services to obtain verifiable parental consent before collecting personal information from children under 13, with limited exceptions. Rather than seek that consent, many services simply prohibit children under 13 from signing up, at least in theory. The law is primarily about protecting children’s privacy, not shielding them from harmful content or other online risks.
But that age restriction makes sense for today’s social media apps. In addition to privacy concerns, there is plenty of content unsuitable for young children, as well as risks they may not have the maturity to protect themselves against, including cyberbullying, excessive screen time, sexual exploitation, misinformation, potential impacts on their well-being and a never-ending parade of scams that ensnare people of all ages.
Age verification then and nowFor most of the 28 years since COPPA was signed into law by President Bill Clinton, online services determined a user’s age simply by asking for their date of birth, with little or no effective verification. In 2008, when I served on the Internet Safety Technical Task Force convened by Harvard University’s Berkman Center for Internet & Society, it was widely recognized that such measures were ineffective at keeping younger children off social media. The task force spent nearly a year examining proposed online safety technologies, including age verification systems, and concluded that none offered a solution we could endorse because of limitations involving accuracy, effectiveness and potential privacy and legal risks.
It took a while, but there are now some reasonably effective technologies that can estimate a user’s age. They can often determine whether someone is over or under 13, 18 or another age threshold. These include AI-powered facial age estimation, which analyzes a selfie without necessarily identifying the person; systems that look for clues in profiles, posts, photos and online behavior; and services that verify age using identification, financial records or digital IDs. But estimates can be wrong, especially near age thresholds, and people challenging incorrect estimates may be asked for government ID or other personal information, raising privacy and security concerns.
The importance of age assurance was underscored by a landmark settlement announced in August between Meta, which operates Facebook and Instagram, and a bipartisan coalition of state attorneys general. The proposed agreement, subject to court approval, requires Meta to strengthen efforts to identify and remove children under 13 and provide additional teen protections. It also includes default daily time limits, nighttime restrictions, limits on notifications during school hours and an option to turn off algorithmically personalized feeds. Nearly all of these limits would require effective age assurance.
Social media isn’t what it used to beAge assurance is arguably more important now than when COPPA was enacted. Since then, algorithms have transformed social media from a place where people primarily interacted with friends and family into platforms that constantly feed users content designed to keep them engaged. That includes posts from so-called “influencers” who make money promoting products, candidates, ideologies and lifestyles, as well as content designed to provoke emotional reactions, sometimes to engage and enrage, regardless of accuracy, quality or impact.
Researcher danah boyd, who served with me on that Harvard task force 18 years ago, was, like me at the time, a strong defender of social media’s potential to bring people together. But in an April 2026 essay titled “Social Media Is Now Parasocial Media,” she argued that its fundamental nature has changed. “The ‘social’ in ‘social media’ has slowly become a misnomer,” she wrote. boyd argues that we’ve entered an era of “parasocial media,” where people develop one-sided relationships with online personalities.
Although social media can still bring people together, it can also drive them apart, as we see from divisive posts, videos and comments on Facebook, X and other platforms. Even Nextdoor, designed to help neighbors communicate, can become a battleground over everything from politics to barking dogs and parking spaces. Instead of the friendly neighbors of Mayberry in The Andy Griffith Show, some discussions resemble the legendary feud between the Hatfields and McCoys.
AI adds another challengeAnother consequential change is generative AI, which became widely available to the public in late 2022 with ChatGPT’s launch. The same 2026 Common Sense Media study found that 49% of tweens ages 9 to 12 and 77% of teens ages 13 to 17 have used AI chatbots or companions at least once, while 26% of tweens and 47% of teens use them at least weekly. Even among 9-year-olds, 36% had tried these tools, despite most major AI chatbot services requiring users to be at least 13.
While I believe AI can have positive uses even for relatively young children, age-appropriate guardrails are important. AI systems can make mistakes, provide misleading information and lead users down dangerous rabbit holes. Unlike traditional search engines, AI chatbots can engage in seemingly personal conversations, offering advice about mental health, relationships and personal safety. Younger children may lack the maturity to recognize when that advice is inappropriate or dangerous.
Age assurance is especially important because AI companies say they increasingly tailor their services based on age. OpenAI’s ChatGPT, Google’s Gemini and Microsoft’s Copilot offer additional safeguards for younger users, including stronger content filters, restrictions on sensitive conversations and, in some cases, parental controls. These protections aren’t foolproof, but they can help make AI more appropriate for younger users without necessarily denying them its benefits.
Advice for parents & caregiversRegardless of laws or company policies, the most important authority when it comes to children and teens is their own parents. My advice is to guide teens on appropriate use of AI and social media through regular conversations that start with questions rather than lectures. Listen to their experiences and concerns, and work together to establish reasonable boundaries.
Although I recognize that good parents may disagree with me, I’m not a fan of letting preteens use social media. I do, however, think many younger children can benefit from AI, provided appropriate guardrails are in place and parents remain involved. That means knowing which services they use and encouraging them to question what AI tells them rather than automatically trusting it.
No law, age-verification technology or company policy can replace good parenting. The goal isn’t simply to keep children away from technology, but to help them develop the judgment and skills to use it safely, responsibly and to their benefit.
Disclosure: Larry Magid is CEO of ConnectSafely, a non-profit internet safety organization that receives financial support from OpenAI, Google and Meta, which are mentioned in this article
——————————————————–
Click Here For The Original Source.
