As legislators across multiple U.S. states move to mandate health warnings on social media platforms, a critical practical question has gone largely unanswered: which warning designs actually work? The answer has real consequences for the mental health of roughly 95% of American teenagers who use these platforms daily, many of whom report measurable impacts on mood, body image, and sleep.
This within-participants randomized clinical trial enrolled 1,012 U.S. participants aged 13 to 29 (mean age 19.3 years; 50% female) in December 2025. Participants were presented with 15 distinct messages in randomized order — 13 topic-specific health warnings covering harms such as depression, anxiety, and negative body image, alongside a screen-time break message modeled on voluntary platform prompts and a neutral control. All messages were embedded within a simulated social media interface to maximize ecological validity. Participants rated each message on a 1–5 scale for two outcomes: perceived effectiveness at discouraging use (primary) and awareness of social media harms (secondary). Linear mixed models quantified average differential effects across message types.
The findings arrive at a pivotal regulatory moment. While the excerpt does not detail which specific warning topics performed best, the experimental architecture is notably rigorous: within-participants design controls for individual variation, the mock social media context avoids abstract hypothetical framing, and the sample spans both minors and young adults — a demographic range rarely captured in a single trial. What makes this incrementally important, rather than paradigm-shifting, is that perceived effectiveness in a controlled setting is a notoriously weak proxy for actual behavior change. Cigarette warning label research offers a cautionary parallel: even highly rated warnings produced modest real-world reductions in use without complementary policy levers. The trial's youth-specific focus is its strongest asset, since adolescent risk cognition differs structurally from adults. Policymakers will need to weigh these perception-based findings carefully before assuming warning labels alone can meaningfully reduce platform engagement or downstream mental health harm.