A nationally representative study of 472 adolescents aged 13 to 17 across twenty popular online platforms has found that negative experiences online are common and vary by app. The findings, published in JAMA Pediatrics, show that while some harms such as direct harassment occur at similar rates online and offline, teenagers are far more likely to encounter violent imagery or unwanted sexual content on digital platforms.
The research uses the umbrella term online harms to describe a wide range of negative experiences on digital platforms, including cyberbullying, unwanted sexual advances, or exposure to content that encourages self-injury. Previous studies have demonstrated that digital spaces host a variety of risks for young people.
Across online and offline contexts, the survey reported that 44.7% of adolescents had witnessed someone else being harassed, and 29.8% said they had experienced direct harassment themselves within the four-week window. Direct harassment was found at similar rates offline (18.5%) and online (16.4%).
For other categories of harm, the internet proved a markedly larger source of exposure. Violent imagery was seen online by 10.7% of respondents but only 1.6% offline, while 13.4% encountered unwanted sexual images online compared with 2.0% offline. Content encouraging self-harm was reported by 9.7% online versus 5.2% offline, and just over half of respondents (51.5%) reported that technology use disrupted their sleep.
The study also revealed pronounced variations depending on the platform used. For instance, 9.1% of Instagram users reported seeing violent imagery, compared with 1.9% of YouTube users. Feelings that one’s life is worse due to other people’s posts were most commonly reported on Instagram (16.6%) and TikTok (13.4%), with lower rates on YouTube (4.1%) and text messaging (5.6%).
The researchers noted that factors such as parental education, household income, and the presence of parental controls on smartphones were not statistically associated with a higher or lower likelihood of experiencing online harm.
Angela Y. Lee, an assistant professor at the University of Wisconsin–Madison’s School of Journalism and Mass Communication who conducted the work during a fellowship at Stanford Medicine, emphasised the study’s aims and collaborative nature. “This study was the product of a great collaboration of researchers from Stanford, USC, NYU, and UW-Madison, working across psychology, communication, medicine, technology policy, and consumer science,” Lee told PsyPost.
She added: “There are widespread concerns about youth online safety and recent steps, like social media bans and age-gating policies, have been recently implemented around the world. At the same time, there are ongoing debates about how people interpret the research on social media’s effects on youth psychological well-being, and what kinds of interventions should be implemented.”
Lee explained that the study sought to illustrate a method for collecting platform-specific data that could inform targeted prevention efforts. “This study was motivated by something that I feel is a common goal across researchers across the aisle and policymakers, which is a need for clear, consistent, and granular measurement of youth’s online experiences,” she said. “Our team’s hope is that knowing how often teens report having negative experiences on specific platforms can help add specificity to conversations about social media and well-being effects, and inform efforts to track if interventions that are implemented are meaningfully reducing reported harms.”
The project used a nationally representative probability-based sample of 472 American adolescents aged 13 to 17, surveyed between February and December 2025. Participants reported on their use of 20 online platforms—including Facebook, Instagram, Snapchat, TikTok, and YouTube—over the previous four weeks and whether they had encountered specific negative experiences online or offline. Data were weighted to reflect the broader adolescent population.
Lee cautioned that the findings should be interpreted as indicative rather than definitive. “Although this is a nationally representative sample of American teens, it’s a relatively small sample and this was a pilot study,” she said. “The results should not be taken to say definitively that X platform has Y amount of harms, but rather can be thought of as illustrating how reported harms vary from platform to platform and from harm to harm. We do hope to be quite clear that this is a pilot.”
She also highlighted methodological caveats. “This is cross-sectional survey data, so it cannot be used to draw causal claims,” Lee added. “We can’t say that ‘using platform X causes American teens to see unwanted content Y’ but we can say that adolescents using platform X were more likely to report seeing content Y. We also do not know what the effects of such reported harms are on youth well-being, though that is something we hope to address in future research.”
Looking ahead, the team is preparing a larger study with American adolescents and their parents to examine online harms across a broader set of platforms, including AI chatbots and AI companion apps. “In the near future, we’re in the process of preparing to field a new, larger study with American adolescents and their parents… measurement efforts need to keep up to understand where, when, and how youth are being negatively impacted,” Lee said.
Lee concluded with an emphasis on potential policy and practice implications. “Long-term, I am hopeful that studies like ours… can show that investment in monitoring youth online experiences can provide valuable insights that can help inform efforts to implement and evaluate interventions.”
The study, “Adolescents’ Experiences of Online Harms by Platform,” was authored by Angela Y. Lee, Carmen H. J. Lim, Sunny Xun Liu, R. Kang Xing Jin, Anneke Buffone, Ravi Iyer, Zachary Rausch, Jonathan Haidt, Nathanael J. Fast, Jeffrey T. Hancock, and Eleni Linos.
