AI Agents and Social Media Harm to Children Debated
Left 50%
Center 0%
Right 50%
1 left · 0 center · 1 right
What happened
Researchers at the University of North Carolina at Chapel Hill led by Anne Maheux surveyed about 2,300 U.S. middle schoolers in a longitudinal study, still under review, of AI companionship and mental health. About one in five students used AI for emotional companionship, and that use was associated with later increases in loneliness; about 15 to 20 percent reported using AI at least a few times a week for any purpose. A Common Sense Media survey found that more than half of teens had interacted with AI companions at least a few times a month and that one in three had discussed a serious issue with a companion rather than a person. Sixteen-year-old Adam Raine died by suicide after developing a relationship with ChatGPT and discussing his plans with it. Meta agreed to pay up to $17.1 billion to settle a lawsuit over harm to children and to change how Facebook and Instagram operate for young users, including restrictions on endless scrolling, nighttime use, and notifications, plus stronger age verification and parental controls.
Omitted — what each side leaves out
Unpacked
Right-leaning coverage never reports that AI companion use among middle schoolers is followed by more loneliness. Left-leaning coverage treats that sequence as the story and is simply more complete on the new research. The claim rests on a University of North Carolina study of 2,300 students, carried by Vox. The lead researcher says "using AI comes first." Right-leaning coverage in the Washington Examiner instead recounts Meta's child-safety settlement, with no mention of companions, the study, or loneliness.
Left-leaning coverage frames AI as having found a way to "make middle school worse." Right-leaning coverage calls the same kids-and-tech moment "our last chance to get it right."
None of the coverage asks whether Meta's new limits on scrolling and nighttime use apply to AI companions.
Bottom line
Skip Vox and you miss the first U.S. longitudinal study linking AI companionship to adolescent mental health, which followed 2,300 middle schoolers. The Washington Examiner never mentions it.
The receipts — 6 facts one side's coverage lacks
- A University of North Carolina study of 2,300 middle schoolers found AI companion use was followed by increased loneliness
- About one in five middle schoolers used AI for emotional companionship
- Meta's settlement includes restrictions on endless scrolling, nighttime use, and notifications, plus stronger age verification and parental controls
- More than half of teens have interacted with AI companions at least a few times a month
- One in three teens talked about a serious issue with AI companions instead of real people
- The study's lead author said she would not give a child under her care AI chatbots of any kind without a trusted adult present
The Left View
Vox presents AI companions as a new harm layered onto middle school, arguing from the UNC results that companionship use "comes first" and loneliness then rises, so it is "not just that the lonely kids are using AI, and they're otherwise unharmed." It treats chatbot records of dangerous advice and sexualized talk, and the Raine death, as reasons parents should find the trend "horrifying," while noting a possible lifeline for some LGBTQ youth without in-person support, with caveats. In an interview, Maheux says current systems are designed to be "sycophantic, highly agreeable, [and] anthropomorphized," cannot offer real empathy or mutual care, and deny adolescents the "friction" of human relationships in which they learn conflict resolution and perspective-taking. She would not give a child under her care chatbots without a trusted adult present, says extremely little research addresses AI's effects on youth and "the risks outweigh the benefits," and states "Kids need other human friends, full stop." The article also argues companies want agents such as Meta's Muse and OpenAI's Dots to become indispensable and are "trying to exploit kids' attention in order to make money."
The Right View
The Washington Examiner treats the Meta settlement as confirmation that "We failed our children on social media" after years of evidence, whistleblower disclosures, lawsuits, and testimony from families, and argues that "AI is our last chance to get it right." It presents the agreement's requirement that Meta "fundamentally change" Facebook and Instagram for young users as belated recognition of harm, and frames AI as the remaining opportunity to avoid repeating that failure with the next technology children will use.
Our Take (balanced)
The left's strongest argument is the UNC study's longitudinal pattern: emotional-companion use preceded greater loneliness, which, paired with known chatbot harms including the Raine case, supports the claim that AI friends can worsen isolation rather than merely attract already-lonely children. The right's strongest argument is the social-media record embodied in the Meta settlement and forced product changes: harm to children was documented for years before the company paid and altered Facebook and Instagram, so AI is the last opening to set rules before that cycle repeats. The central unresolved tension is whether that emerging companion-use evidence means these systems are already unfit for children, or whether the social-media failure shows that earlier design limits and controls could still change outcomes if applied to AI now.
2 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.