The Dark Side of Social Media

What Happens When Algorithms Choose What Young People See

Social media platforms run on attention. The longer someone scrolls, the more ads they see, and the more revenue the platform generates. To maximise that attention, algorithms track every interaction — likes, shares, how long you pause on a video — and serve up whatever keeps you engaged.

Violent content happens to be extraordinarily effective at this. A fight video, a threatening confrontation, or graphic footage triggers a strong emotional reaction — fear, outrage, morbid curiosity — and that reaction translates directly into engagement metrics. According to research from the Youth Endowment Fund, 70% of teenagers have encountered real-life violence on social media, with over half (56%) reporting they've seen footage of physical fights between young people. Over a third have seen content involving weapons.

The critical point is that most young people don't go looking for this material. The algorithm brings it to them. One curious click on an emotionally charged video can trigger a cascade of increasingly extreme recommendations — what researchers call a "rabbit hole" effect. A 2025 study published in Frontiers in Psychology found that recommender algorithms actively normalise toxic content by prioritising metrics like shares and reactions, creating feedback loops that amplify fear, outrage, and polarisation.

For young users whose brains are wired to seek novelty and reward, these loops are especially hard to escape.

The Mental Health Toll Is Measurable

The psychological effects of sustained exposure to violent and distressing content are well-documented and increasingly precise.

Anxiety is one of the most consistent findings. When algorithms continuously surface threatening content — fights, weapons, aggressive confrontations — they distort a young person's perception of the world. Everyday environments start to feel more dangerous than they actually are. The U.S. Surgeon General's Advisory on Social Media and Youth Mental Health warned that high levels of social media use predicted double the future risk of depression among adolescents.

Depression follows a similar trajectory. The 2026 World Happiness Report concluded that social media is harming adolescents at a scale large enough to cause changes at the population level — not just in individual cases, but across entire cohorts. Exposure to hate-fuelled or discriminatory violence, which is common in algorithmically curated feeds, predicts sharper rises in depressive episodes, particularly among girls and marginalised youth.

Then there's desensitisation. Repeated exposure to graphic content gradually dulls empathy and blunts emotional responses to real suffering. What once provoked shock becomes background noise. This emotional numbing doesn't just affect how young people react to content on a screen — it shapes how they respond to real people in real situations, weakening the compassionate social bonds that are essential during adolescence.

From Screen to Behaviour

The effects don't stay contained within a feed. Research consistently links exposure to violent social media content with real-world behavioural changes.

Violent content models aggressive behaviour. When confrontation is rewarded with likes, shares, and attention, it teaches young viewers that aggression has social currency. A Frontiers in Psychology study found a significant association between exposure to violent social media content and youth violence in real life. Nearly two-thirds of young people surveyed said social media played a role in escalating conflicts — online arguments spilling into in-person confrontations, comments amplifying disputes, and people saying things digitally that they would never say face-to-face.

Cyberbullying and online harassment, which algorithms can amplify by boosting confrontational content, further blur the line between digital and physical aggression. In vulnerable young people, this manifests as heightened irritability, risk-taking, and coercive patterns in peer relationships. Over time, the normalisation of toxicity contributes to a culture where empathy erodes and aggression becomes a default response — precisely during the years when identity and social skills are still forming.

Why Adolescent Brains Are Especially Vulnerable

Adolescence is a period of intense brain development, and this is precisely what makes young people so susceptible to the effects of algorithmically served violent content.

The brain's emotional centres — the amygdala (which processes fear and threat) and the nucleus accumbens (the reward centre) — are highly active during the teenage years. Meanwhile, the prefrontal cortex, which handles judgement, impulse control, and emotional regulation, is still developing and won't fully mature until the mid-twenties. This creates what neuroscientists describe as a developmental imbalance: the emotional accelerator is fully engaged while the cognitive brakes are still being installed.

Research from Columbia University has shown that habitual social media checking preferentially affects the amygdala and the dorsolateral prefrontal cortex — the very regions responsible for emotional processing and reasoned decision-making. Accelerated cortical thinning of the prefrontal cortex has been associated with higher levels of depression and diminished impulse control.

The constant adrenaline spikes from threatening content essentially mimic chronic stress on the developing brain. Over time, this can alter neural pathways tied to fear processing and emotional regulation, potentially creating lasting vulnerabilities — heightened anxiety disorders, reduced resilience to adversity, and impaired capacity for empathy.

What You Can Actually Do About It

Understanding the problem is the first step. Here's how to act on that understanding.

Start with awareness, not panic. The goal isn't to ban screens or demonise technology. It's to understand that algorithms are designed to maximise engagement, not wellbeing — and to make informed decisions based on that reality.

Audit the feed. If you're a parent or caregiver, sit with the young person in your life and scroll through their feed together. Not as surveillance, but as a conversation starter. What does the algorithm think they want to see? Is that actually what they want?

Use platform controls strategically. Most platforms offer content filtering, restricted modes, and the ability to mark content as "not interested." These tools aren't perfect, but they do influence what the algorithm serves next. Actively training the algorithm to deprioritise violent content makes a measurable difference.

Build critical media literacy. Help young people understand why they're seeing what they're seeing. When they recognise that a shocking video appeared in their feed because it drives engagement — not because it reflects the real world — it loses some of its psychological power.

Support legislative momentum. The Kids Online Safety Act, reintroduced in 2025 with bipartisan support, would require platforms to exercise reasonable care in preventing harm to minors and provide meaningful parental controls. Advocacy for these measures sends a signal that platform accountability matters.

Prioritise offline connection. The most robust protective factor against social media harms remains strong, trust-based relationships in the physical world. Regular face-to-face time with family and peers provides the emotional grounding that no algorithm can replicate.

Key Takeaways

• Algorithms prioritise violent content because it drives engagement, and 70% of teenagers encounter real-life violence in their social media feeds without seeking it out.
• Sustained exposure is linked to increased anxiety, depression, desensitisation, and aggressive behaviour — effects documented at the population level.
• Adolescent brains are especially vulnerable because emotional processing centres mature faster than the regions responsible for impulse control and judgement.
• Practical steps — auditing feeds, using platform controls, building media literacy, and maintaining offline relationships — can meaningfully reduce harm.
• Legislation like the Kids Online Safety Act represents a growing recognition that platform design, not just individual behaviour, needs to change.