Mental health warning pop-ups on social media more persuasive for young users, study finds

A new US experiment reveals that mental-health warnings displayed as pop-up labels on social media platforms significantly influence young users’ perceptions, with potential implications amid ongoing legislative efforts to introduce warning labels and address industry resistance.

Mental-health warnings shown as pop-up labels on social media apps were the most persuasive messages in a new US experiment, with young users saying they would be more likely to cut back after seeing them than after either a neutral notice or a softer prompt to take a break. The randomised trial, published in JAMA Health Forum on 4 September 2026, tested the idea on 1,012 people aged 13 to 29 and was released against a much tougher political backdrop for platforms, days after Meta agreed to a multistate settlement of up to $18 billion over claims that its products harmed younger users. (jamanetwork.com)

According to the journal paper, the researchers recruited participants in December 2025 through the InnovateMR panel and split the sample almost evenly between teenagers aged 13 to 17 and young adults aged 18 to 29. Participants were shown 15 messages in random order on screens designed to resemble Instagram: 12 researcher-developed health warnings covering six themes, California’s proposed warning, a voluntary screen-time-break reminder similar to one used by TikTok, and a seatbelt-safety message used as the control. The registered trial, listed by Stanford as no longer recruiting, also identified a secondary aim of comparing warnings refined with artificial intelligence against versions that were not. (jamanetwork.com)

Every warning category outperformed the neutral control on the two measures the researchers cared about most: whether a label discouraged social media use and whether it made harms more visible. The stronger health warnings also beat the voluntary break reminder on both counts, with one notable exception: an addiction-focused message did not clearly do better than the softer prompt. Warnings about overall mental-health harms, and those referring specifically to depression and anxiety, received the highest ratings, reinforcing the study’s central finding that broad well-being messages may land harder with young audiences than reminders about screen time alone. (jamanetwork.com)

The wording mattered. Stanford Medicine said one tested message on sleep read, “SURGEON GENERAL’S WARNING: Using social media causes sleep problems. This is bad for your health.” Another warning told users that social media “is associated with significant mental health harms for children and young adults”. Anna Grummon, the study’s lead author and an assistant professor of paediatrics, said the field had lacked evidence on both content and effect: “We haven’t known what warnings for social media should say, or if they’re promising for changing behavior.” She added that “Sometimes, subtle changes to warnings really matter.” Asked why the mental-health warnings did best, Grummon told News-Medical that they were probably harms people felt they recognised in themselves or friends. (med.stanford.edu)

The study sits inside a fast-moving policy debate that has advanced well beyond the thought experiment stage. Stanford’s account says the research was inspired by Vivek Murthy’s 2024 call, while he was US surgeon general, for warning labels on social media similar to those used on tobacco and alcohol. Four states , California, Colorado, Minnesota and New York , have since passed laws requiring warnings in some form, and California’s law is due to take effect on 1 January 2027. Grummon also noted that a bipartisan coalition of 42 state attorneys general later urged Congress to enact a federal version, while lawmakers in Texas, Ohio and Pennsylvania have introduced related proposals. (med.stanford.edu)

That legislative push is already colliding with legal objections from the industry. Stanford reported that California’s own label, though longer and more legally cautious than the researchers’ preferred versions, performed roughly as well as the other health warnings. Thomas Robinson, the study’s senior author, said of the state wording: “It’s not the message we would have designed,” but added that the team understood why it performed well and that it had been drafted to survive court scrutiny. He also said that “All these bills are in litigation, being challenged by the social media industry.” The study’s publication so soon after the Meta settlement gives the findings extra political weight, because the case centred on whether major platforms deliberately built products that held children’s attention at the expense of their mental health. (med.stanford.edu)

The authors are careful not to claim they have proved pop-ups will change real-world behaviour. What they measured was perceived effectiveness: participants answered, among other questions, “How much does this message discourage you from wanting to use social media?” and “How much does this message increase your awareness of the harms of using social media?” The JAMA paper says those perceptions are often linked to actual effects, but it calls for longer trials or quasi-experiments that track what people really do after repeated exposure. Grummon drew another distinction from tobacco-style labelling, arguing that social media is “a totally different product” because the warnings are not printed on packaging but arrive as interruptions inside the experience itself. (news-medical.net)