Social media algorithms are harmful to adolescent mental health

What's this about?

People disagree about whether social media rules (algorithms) can hurt teens’ mental health. The rules choose which posts people see, often to keep them online.

What supporters say

• Heavy or hard-to-stop social media use links with more sadness, worry, and lower well-being.
• Algorithmic feeds may push teens toward more comparing, body-image stress, and hard-to-stop use.
• These feeds may show harmful posts again and again, especially to at-risk teens.

What critics say

• Most studies look at social media use, not the computer rules that choose the posts.
• Social media affects teens in very different ways, based on the teen and the activity.

How to read this

The number of points on each side does not show which side is right; strong evidence matters more.

The bottom line

The evidence shows real risks from heavy social media use, and algorithms may make some risks worse. But we are not sure yet whether algorithms alone cause widespread harm.

The claim is that social-media recommendation systems can harm adolescent mental health by repeatedly promoting damaging content, encouraging comparison and driving compulsive use. The evidence raises serious concerns, but it does not yet show that algorithms cause widespread harm independently of social-media use itself.

The case for

The strongest argument is about how recommendation systems may work. Platforms designed to maximize engagement can repeatedly direct vulnerable adolescents toward harmful, appearance-focused or emotionally intense material. That exposure may reinforce further viewing and make compulsive use harder to stop. Research on adolescents’ experiences, along with analyses of engagement-focused design, supports this as a plausible pathway rather than a proven explanation for all users.

The U.S. Surgeon General has identified harmful content, social comparison, cyberbullying and sleep disruption as risks in young people’s social-media environments. While that evidence does not show that algorithms alone cause these problems, recommendation systems may help determine which material users see repeatedly. Appearance-based social-media use is one of the clearest specific concerns: prospective research has linked it with later measures of adolescent mental health, supporting a possible connection through comparison and body-image pressure (see Figure 3).

There is also stronger evidence about intensive or problematic social-media use more generally. Reviews and long-term studies associate heavy or difficult-to-control use with depression, anxiety, internalizing symptoms and lower well-being. A randomized trial found that reducing screen time improved some mental-health outcomes (see Figure 4). These findings suggest that intense digital engagement can affect well-being, and they are consistent with concerns that systems encouraging repeated use could contribute to harm. Algorithms may intensify risks that are already associated with problematic engagement 23.

The concern is likely greatest for particular groups and situations. Adolescents with existing vulnerabilities may be more affected by harmful content, social comparison or compulsive engagement. Platform-specific reporting has also raised concerns about internal research into these risks, although such reporting is not the same as a peer-reviewed causal study.

The case against

The main problem is attribution. Much of the strongest research measures time spent on social media, overall use or problematic engagement—not how recommendation algorithms rank and deliver content. Reviews repeatedly note that studies rarely isolate algorithms from the content, social setting and broader experience of using a platform 4.

Even the randomized screen-time trial reduced overall smartphone exposure. It therefore cannot show that recommendation systems, rather than other forms of phone use, caused the improvements. Researchers also face limited access to platform data, changing algorithms, measurement problems and the possibility that poor mental health leads to heavier social-media use, rather than the reverse.

Effects also differ substantially between adolescents. Young people use social media for different activities and in different social circumstances. Active, supportive or socially connected use may be neutral or beneficial, including by providing friendship, identity exploration, information and peer support. These benefits may be especially important for some marginalized adolescents. Evidence therefore does not support the idea that every adolescent is harmed by algorithmic feeds 5.

The bottom line

The evidence supports a qualified warning, not a universal verdict. Recommendation systems are plausibly harmful for some adolescents, especially when they amplify harmful content, appearance-based comparison or compulsive engagement 12. The broader link between problematic social-media use and poorer mental health is supported by reviews, longitudinal studies and an intervention, so the issue warrants scrutiny, transparency and targeted protections.

But confidence is lower that algorithms themselves are the independent cause. Existing research more reliably identifies associations, possible mechanisms and vulnerable groups than it measures the separate effect of ranking systems. The evidence therefore favours serious, targeted concern moderately strongly, while falling short of showing a general algorithm-specific harm across all adolescents. The central unanswered question is whether algorithms worsen mental-health outcomes on their own, amplify existing vulnerabilities, or mainly pass along harms created by content, social context and overall use.