Warning messages naming depression, anxiety, and overall mental health harms were rated as more discouraging by teenagers and young adults than the take-a-break prompts social media companies already display, according to a randomized experiment published this week in JAMA Health Forum.
The finding lands as four states prepare to require warnings on platforms. California, Colorado, Minnesota, and New York have passed laws, and California's takes effect on January 1, 2027. Texas, Ohio, and Pennsylvania have proposed legislation. Researchers said the wording had not previously been tested rigorously on the people it targets.
What the study measured, though, is what a young person predicted about their own behavior, not what they actually did afterward. That gap is the most important thing for a parent to hold onto before drawing conclusions about screen time at home.
The Design and the Comparison That Mattered
Stanford Medicine researchers administered questionnaires to 1,012 young people, about half aged 13 to 17 and about half aged 18 to 29. Each participant rated 14 warnings and one control message, presented in random order and formatted to look as though they were appearing on an Instagram page.
The warnings covered health topics including social media addiction, negative body image, sleep disruption, depression, anxiety, and overall mental health harms, plus a message noting that social media has not been proven safe for young users. Researchers wrote two warnings per topic. They also tested California's statutory language and a screen time break message similar to one TikTok uses. The control message discussed seatbelt safety, relevant to young people but unrelated to social media.
Participants rated each message on two questions: how much it discouraged them from wanting to use social media, and how much it increased their awareness of social media harms.
Every warning outrated the seatbelt control. More telling, every warning mentioning a health effect, except the addiction warning, outrated the screen time break message. The messages rated most effective concerned overall mental health harms, depression, and anxiety.
Perceived Effectiveness Is Not Measured Behavior
This is an experiment measuring perceived message effectiveness. It is an established approach in warning label research, and it is not a behavioral trial. No one's actual screen time was logged.
Lead author Anna Grummon, an assistant professor of pediatrics, was direct about the limits. "We haven't known what warnings for social media should say," she said. She also noted that the data cannot explain why the mental health messages performed best, offering only a hypothesis that those harms resonate because young people can picture them happening to themselves or their friends.
Grummon drew a distinction that matters for anyone reading this as settled policy evidence. With tobacco and alcohol, she said, well-designed warning labels change behavior on average, but social media is a different product because the state laws are written so the warnings function as pop-ups rather than labels that live permanently on a package. How that format affects behavior remains unstudied.
Senior author Thomas Robinson said teens and young adults consistently reported that the tested messages would reduce their use compared with the control. The researchers had expected the two age groups to respond differently and found they did not.
The study was funded by the Stanford Maternal and Child Health Research Institute and the Department of Pediatrics at the Stanford University School of Medicine, with contributions from researchers at the University of North Carolina and the University of California, Davis.
The Policy Fight Running Alongside the Science
California's warning language is longer and more complex than what the researchers would have written, and it still performed on par with the other health warnings. Robinson said the bill's authors explained their reasoning, and that the wording was designed to withstand legal scrutiny. "It's not the message we would have designed," he said.
That detail points to the real constraint. All of these state laws are in litigation, challenged by the social media industry. A warning that reads awkwardly but survives a First Amendment challenge delivers more public health value than an elegant one that gets struck down.
The broader push began with a 2024 call from then Surgeon General Vivek Murthy to add health warnings to social media in the way tobacco and alcohol carry them. A bipartisan coalition of 42 state attorneys general subsequently petitioned Congress to enact warnings at the federal level.
Separately, a federal trial over social media design and teen mental health opened in Oakland, a case turning on product design rather than disclosure. A warning label requirement would not resolve the design questions at issue there.
A Reasonable Takeaway for Parents
Nothing here supports a specific screen time rule for a specific household. The study does not establish that warnings reduce use, does not measure mental health outcomes, and does not identify a safe amount of social media use.
What it does offer is a conversational finding. When young people were asked which framings landed, they picked the ones naming depression, anxiety, and mental health rather than time management. A parent who has been framing the issue as hours spent might find that a conversation about how a platform makes a teenager feel goes further than one about a screen time limit.
Warning labels are one tool among many, as the researchers emphasized. Any parent concerned that a teenager's mood, sleep, or eating has changed should raise it with a pediatrician or a mental health professional rather than treating social media as the sole explanation. Excessive use has been linked to depression, anxiety, worsened body image, and sleep disruption in prior research, but those are associations rather than proof of cause in any individual case.
The researchers' next step is testing whether warnings actually reduce use. California's law takes effect on January 1, 2027, creating the first large-scale natural experiment. The confirmed finding is that message content mattered, and mental health framing tested strongest. The people most affected are teenagers in the four states with laws on the books. The reasonable family action is a conversation, not a policy change. The central uncertainty is whether any of these messages change behavior at all.
Key Questions Answered
What did the study find? Among 1,012 teenagers and young adults, warnings naming overall mental health harms, depression, and anxiety were rated most discouraging. All health warnings except the addiction warning outrated a screen time break message similar to one platform already uses.
Did the study measure actual social media use? No. It measured how participants rated messages on discouraging use and raising awareness. No behavior was logged, and the researchers said behavioral testing is the necessary next step.
Who took part? About 1,012 people, roughly half aged 13 to 17 and half aged 18 to 29. Participants rated 14 warnings and one control message about seatbelt safety, in random order, formatted like an Instagram page.
Which states require social media warnings? California, Colorado, Minnesota, and New York have passed laws. California's takes effect January 1, 2027. Texas, Ohio, and Pennsylvania have proposed legislation. All the laws face industry litigation.
How did California's official wording perform? It performed on par with the other health warnings despite being longer and more complex. A study author said the language was written to withstand legal scrutiny.
Who funded the research? The Stanford Maternal and Child Health Research Institute and the Stanford University School of Medicine Department of Pediatrics.
What should a parent do with this information? Treat it as a framing insight rather than a rule. If a teenager's mood, sleep, or appetite has changed, raise it with a pediatrician or mental health professional rather than assuming social media is the sole cause.
