Harmful Content and the Algorithm: Protecting a Vulnerable Teen From What the Feed Delivers

What doomscrolling really is

The word arrived during the early pandemic, when millions of adults found themselves unable to stop refreshing a stream of frightening news at two in the morning. It described something most people recognized immediately, and it quickly settled into ordinary conversation as a mild, slightly self-mocking confession — the digital equivalent of eating a whole packet of biscuits. Filed under bad habits, somewhere near procrastination.

That framing is comfortable, and it is part of the problem. It treats the behavior as a small failure of self-discipline, which means the cure is presumed to be willpower: just put the phone down. Applied to a teenager, it becomes a daily argument about hours and a chore chart, and it quietly assigns the blame to the child. But that diagnosis is incomplete, and the cure it implies — willpower — misses the real mechanism almost entirely.

Doomscrolling is the compulsive, hard-to-interrupt consumption of a continuous stream of distressing, negative, or emotionally charged online content, sustained well past the point where it makes the viewer feel worse. The key word is compulsive. A teenager who cannot stop is not weak-willed. They are doing exactly what the product in their hand was engineered to produce. The feed has no natural end, no last page, no closing credits — it is built precisely so that the question "should I stop now?" never has an obvious moment to be asked. And this is not a fringe behavior at the edges of teenage life. The Pew Research Center has found that nearly half of US teenagers now say they are online almost constantly — roughly double the share of a decade ago — so for many teens the feed is not an occasional visitor to the day but its background hum.

And there is a deeper point that the willpower framing hides entirely. The harm is not really in the scrolling. It is in what is being scrolled. Two teenagers can each spend the same two hours on the same app and have completely different experiences, because each one is shown a different feed — and the feed is not assembled by chance, by their friends, or by them. It is assembled by an algorithm optimizing for something specific. This guide is about that algorithm: what it is choosing, why it sometimes chooses badly for a vulnerable teen, and what a parent can actually do about it.

How the algorithm decides what your teen sees

For most of the internet’s history, a feed was a list. You followed people, and you saw what they posted, in the order they posted it. When you reached something you had already seen, you stopped, because there was nothing new. That design is almost gone. The feeds where teenagers now spend their time are recommendation algorithms — systems that do not show what a teen’s friends posted in order, but rank an effectively infinite supply of content and serve whatever the system predicts will hold that specific teenager’s attention next.

The prediction is built from data the teen produces without noticing. Not just the obvious signals — likes, follows, shares — but the quiet ones: how long they lingered on a clip before swiping, whether they watched it twice, whether they slowed down, whether they turned the sound on, what they searched at midnight, what they watched to the end. A modern feed reads hesitation. A teenager who pauses for three extra seconds on a sad video has, without intending to, told the system something, and the system files it away.

None of this would matter if the algorithm were optimizing for the teenager’s wellbeing. It is not. It is optimizing for engagement — time, attention, sessions, returns — because those are the numbers the platform is built around and can measure. And here is the mechanism every parent needs to hold onto: engagement and wellbeing are not the same thing, and content that provokes strong emotion is unusually engaging. Outrage, fear, envy, shock, and sadness all hold attention extremely well. A calm, balanced, reassuring video competes badly against a clip engineered to make a viewer anxious. The algorithm is not malicious — it has no concept of harm at all. It is simply an optimizer that has discovered, across billions of sessions, that distress performs.

Two features sharpen the effect. The feed is infinite: autoplay and endless scroll remove every natural stopping point, so disengaging requires an act of will at a moment the design has worked to make frictionless. And the feed is personalized to a degree that has no precedent: your teen’s feed is not a slightly tuned version of a shared feed, it is unique to them, shaped session by session toward whatever their own behavior reveals. That is why you cannot simply look over their shoulder and see "the app." There is no the app. There is only their feed, built from their hesitations.

It helps to compare this with the media a parent grew up with. A television channel had a schedule and a visible editor; a newspaper had a masthead and a front page that everyone in the house saw the same way. A parent could evaluate those choices, argue with them, and switch them off. The feed has an editor too — the recommendation algorithm — but that editor is invisible, is accountable to no parent, never publishes its reasoning, and is tuned not to a general audience but to one specific child, in private, around the clock. The shift is not that teenagers now consume more media. It is that the thing deciding what they consume has become both far more powerful and far harder for a parent to see.

Using social media is not inherently beneficial or harmful to young people. Its effects depend in large part on what adolescents are exposed to and engage with, and on each young person’s own strengths and vulnerabilities.

— American Psychological Association, Health Advisory on Social Media Use in Adolescence

The categories of harmful content

"Harmful content" is a vague phrase, and vagueness is unhelpful to a worried parent. It is worth being concrete about what the category actually contains, because the five kinds below behave differently, reach a teen by different routes, and call for different responses. None of this is meant to alarm. Most of what an algorithm serves a teenager is ordinary — friends, music, jokes, hobbies, creators they genuinely admire. The point is not that the feed is a sewer; it is that the same engagement-driven machinery that delivers the ordinary content can, for a vulnerable teen, also deliver the content below, and do so persistently.

Pro-self-harm and pro-eating-disorder content

This is the category that most often frightens parents, and with reason. It is not, in the main, instructional material; it is content that presents self-harm, suicidal feelings, or disordered eating as relatable, even as a kind of belonging — an identity shared by a community that understands. It is also adept at surviving moderation, drifting through a constant churn of coded hashtags and euphemisms a moderation system has not yet learned. Crucially, it tends to find teenagers who are already struggling rather than manufacturing distress in a well teen. A platform’s recommender, noticing that a sad or anxious teenager lingers on this material, treats the lingering as interest and supplies more.

Violent and graphic content

Real footage of violence, accidents, fights, abuse, and cruelty circulates widely, often stripped of any context and surfaced purely because shock keeps people watching. A teenager does not have to go looking for it. The cumulative cost of repeated exposure is rarely a single dramatic reaction; it is quieter — a gradual dulling of the sense of what is ordinary, and a low, persistent hum of background anxiety about how dangerous the world appears to be. A parent rarely sees the clip itself, because it has usually been swiped past long before any conversation happens; what remains visible is the residue, in a teenager who has become harder to reach and quicker to expect the worst.

Extremist and radicalizing content

Material that recruits toward hateful, misogynistic, or extremist worldviews almost never arrives labeled as such. It comes dressed as comedy, as gaming culture, as fitness, as blunt "common sense" about how the world really works, as self-improvement advice for a teenager who feels adrift. The shift in tone is gradual and is the subject of the rabbit-holes section below.

Age-inappropriate and sexual content

Pornographic and sexualized material reaches teenagers who never searched for it — through a recommendation, a reposted clip, a link in a group chat, a message from a stranger. Beyond the immediate shock of unwanted exposure, the steadier concern is that it can quietly shape a young person’s expectations about bodies, sex, and consent well before they have the experience or the conversations to put it in perspective.

Misinformation and health disinformation

The last category is the least dramatic and the easiest to underestimate. Feeds carry a heavy load of false or distorted claims about health, the body, nutrition, science, and current events, and the most persuasive of it is produced with real polish and delivered with total confidence. For a teenager, the health and body strand is the most consequential: confident "advice" about restrictive diets, supplements, fitness extremes, or untested treatments, presented by someone who looks authoritative, can do measurable harm.

Why vulnerable teens are hit hardest

Everything described so far applies to every teenager with a feed. But the effects are not evenly distributed, and understanding why is the most important idea in this guide. The reason is not that some teenagers are weaker. It is that the algorithm and the vulnerability interact — they form a loop, and the loop runs faster the more vulnerable the teenager is.

Recall how the recommender works: it watches behavior and amplifies whatever holds attention. Now consider a teenager who is anxious, low, or depressed. Distressing content tends to hold their attention more — not because they enjoy it, but because a low mood narrows attention toward exactly that kind of material. The teen lingers. The algorithm, which cannot distinguish "this is helping me" from "I cannot look away," reads the lingering as a preference and delivers more. More distressing content deepens the low mood. The deeper mood produces more lingering. This is the engine: vulnerability shapes behavior, behavior trains the algorithm, and the trained algorithm intensifies the vulnerability. A teenager can enter that loop in an ordinary low week and find, a month later, that their feed has quietly reorganized itself around their worst hours.

The same mechanism runs for a teen at risk of an eating disorder, whose attention is drawn to body- and food-related content, and who is then served a feed increasingly dense with it. It runs for a grieving or frightened teen and a feed of doom. In each case the algorithm is not targeting the vulnerability. It is simply doing what it always does — and what it always does happens to be precisely the wrong thing for a struggling teenager.

Consider a composite of the kind clinicians describe often. A fourteen-year-old with an anxiety disorder starts watching ordinary fitness and "what I eat in a day" videos — a healthy interest, nothing she went looking to be harmed by. But she lingers a little longer on the clips about restriction and "clean eating," because anxiety pulls attention toward control, and within a few weeks that corner of the feed has quietly widened to fill it. The U.S. Surgeon General’s advisory reports that 46% of adolescents aged 13 to 17 say social media makes them feel worse about their body image. For a teen already prone to that worry, the algorithm does not invent the vulnerability — it locates it, and then it feeds it.

Neurodivergent teenagers — those on the autism spectrum, with ADHD, or with related differences — can be affected in additional ways. A tendency toward intense, focused interests, which is often a real strength, can also mean a deep dive into a topic goes deeper and is harder to climb back out of. Difficulty disengaging from a screen interacts badly with a feed engineered to have no stopping point. And a literal, trusting reading of content can make confidently delivered misinformation or the framing of extremist material harder to recognize as a pitch. None of this means a neurodivergent teen should be kept offline; for many, online spaces are a genuine and valuable source of connection and community. It means the curation and conversation in the later sections matter more, not less.

We are in the middle of a national youth mental health crisis, and I am concerned that social media is an important driver of that crisis. We cannot conclude that social media is sufficiently safe for children and adolescents.

— U.S. Surgeon General, Advisory on Social Media and Youth Mental Health

It is worth keeping that statement in proportion. The evidence on social media and adolescent mental health is genuinely mixed, and serious researchers disagree about how large the average effect is. What is far less disputed is the point this section makes: averages hide the teenagers who matter most here. A platform can be roughly neutral for a typical, well-supported teen and still be actively harmful for a smaller group of vulnerable ones — and it is that group this guide is written for.

Rabbit holes and radicalization

A rabbit hole is the gradual narrowing of a feed, step by step, from broad mainstream content toward a niche, intense, and sometimes extreme version of it. No single step is alarming, which is exactly what makes the process effective. The teenager is never shown something shocking by a system that has correctly judged they are not ready for it. They are shown something slightly more pointed than the last clip — and then, once that registers as engagement, something slightly more pointed again.

Consider how this looks for a teenage boy who feels lonely, unsure of himself, and behind his peers. He watches some ordinary fitness and self-improvement videos — a completely healthy interest. The algorithm has plenty of adjacent content, and some of it pairs workout advice with a harder edge: a little resentment, a theory about why life is unfair to young men, a confident older voice explaining who is to blame. If those clips hold his attention even slightly better — and content with a grievance often does — the feed leans that way. Weeks later the proportions have shifted. The fitness is now a minority of the feed, and the surrounding worldview has hardened. He did not seek out extremism. He sought out push-ups, and an optimizer did the rest.

The same architecture drives other pipelines — conspiratorial thinking, rigid political extremes, communities organized around contempt for some group. Whether recommendation algorithms cause radicalization, or mainly accelerate teens already drifting that way, is still debated by researchers, and the honest answer is that the feed is an amplifier more than an origin. But amplification is enough to matter. The protective insight for a parent is that you will almost never catch the dramatic moment, because there is no dramatic moment. What you can notice is drift: new vocabulary, a hardening of opinion, contempt arriving where curiosity used to be, a sense that someone online is now explaining the world to your teenager. That is the cue for conversation — calm, curious, and genuinely interested — not for confiscation, which simply ends the conversation and leaves the worldview intact.

Dangerous viral challenges

Few online-safety topics generate more parental fear, or more confusion, than the viral challenge. It deserves a calm and honest treatment, because the panic itself causes harm. Here is the uncomfortable truth: a large share of the most frightening "challenge" stories that circulate through news segments and parent groups are exaggerated, distorted, or entirely fabricated. They spread because alarm is engaging — the same mechanism that drives the rest of this guide. And these scares carry a specific cost: detailed, breathless coverage of a supposed dangerous challenge can introduce it to children who had never heard of it, and frame it as something other kids are doing.

That does not mean the risk is zero. Genuinely dangerous challenges do exist, and some have led to real injuries and deaths — typically those involving choking, suffocation, ingesting harmful substances, or physical recklessness. The risk is real but smaller and more specific than the general atmosphere of panic suggests. Two things make a challenge actually dangerous: a direct physical hazard, and a strong social pull to film it and post the result.

The useful parental response is not to forward every warning that lands in a group chat — that behavior is part of the amplification machine. It is to do two quieter things. First, verify before reacting: check whether a claimed challenge has been confirmed by a credible source rather than a screenshot, because Common Sense Media and similar organizations regularly debunk scares that turn out to be hoaxes. Second, and more durably, give your teenager a single transferable idea instead of a list of forbidden challenges you can never keep current: something going viral is not evidence that it is safe. A teenager who has genuinely absorbed that one principle is protected against next year’s challenge, the one nobody has named yet — which a list of this year’s challenges can never do.

Warning signs you can see

Parents often assume the harm of a feed is invisible by nature — that it all happens inside a device they cannot read. The specific clips may be out of view, but a feed that has turned against a teenager almost always shows itself, for days or weeks, in behavior. The signals are not exotic. They are the ordinary signs of a young person under stress — and what has changed is how often the cause now sits in a recommendation algorithm.

• Mood tied to the feed Anxiety, sadness, anger, or agitation that reliably follows a scrolling session rather than a real-world event.
• Compulsive checking Reaching for the phone the moment it is set down, distress when separated from it, scrolling that visibly is not enjoyable but cannot be stopped.
• Sleep erosion Late-night or all-night use, exhaustion in the morning, a phone that goes to bed with them — feeds are designed to be hardest to leave when a teen is tired.
• A darkening worldview New pessimism about the future, the body, other people, or whole groups, often phrased with a certainty that did not come from your conversations.
• Body and food preoccupation New fixation on appearance, weight, diet, or exercise, or content of that kind appearing on a shared screen.
• New vocabulary or worldview Slang, talking points, or an "explanation of how things really are" that seems to arrive whole, from a source online.
• Withdrawal Pulling away from family, friends, hobbies, and the offline world the feed is steadily out-competing.
• Distress after specific content Mentioning, or visibly reacting to, disturbing things seen online — or going abruptly, smoothly silent about the online part of life.

No single item on that list is, by itself, evidence of anything. Teenagers are entitled to bad moods, to privacy, to intense new interests, and to changing their minds. What matters is clustering: two, three, or four of these appearing together within a short window deserves a calm, careful response. And the response begins with the relationship, not the device. Open with the young person — ask how they are, what has been on their mind, what they have been watching lately — rather than with what you have noticed on a screen. Lead with the device and you teach the lesson the feed already teaches: that adults are a problem to be managed rather than a resource to be used.

A second, quieter class of signal is worth training yourself to notice: the change in pattern rather than the dramatic event. A teen who used to narrate their day and now will not, a previously easy child who becomes uniformly irritable in the hour after putting the phone down, a sudden flattening of interests that used to be theirs — each is worth a gentle, curious question rather than an accusation. The hardest signal of all is the one that looks like nothing: a teenager who has simply gone quiet and still. Silence, in a child who used to have a lot to say, is information.

What parents can do

The single most useful shift a parent can make is to stop thinking about limiting the feed and start thinking about curating it. Time limits still have a place — protecting sleep and homework is worth doing — but a time limit does nothing about what the algorithm delivers inside the time that remains. Curation does. The goal is a feed that carries more of what a teenager genuinely values and less of what is quietly harming them, and most of that is achievable with settings, signals, and conversation.

Start with the platform’s own controls, and do this with your teen rather than behind them. Most major platforms now offer a sensitive-content or content-preferences setting, a teen or restricted account mode with stricter defaults, and tools to clear watch history, mark posts as "not interested," and mute or unfollow accounts. Several also provide family or parental dashboards. After tightening the settings, retrain the feed deliberately: a feed that has drifted somewhere bad will not correct itself, but it does respond quickly to new signals, so a session spent actively engaging with genuinely good content teaches the recommender as effectively as months of harm taught it the other way. Organizations such as Internet Matters publish current, platform-by-platform setup guides, which are more reliable than any list a single article could keep up to date.

Interfaces change often, but the controls worth knowing by name are reasonably stable. As a starting point for the platforms a teenager is most likely to use:

• YouTube Restricted Mode, the supervised-account settings, and the option to clear and pause watch history — one of the strongest signals shaping its recommendations.
• TikTok Family Pairing, Restricted Mode, content keyword filters, and the "refresh your For You feed" option, which clears the slate and lets the recommendations start over.
• Instagram Teen Accounts, which apply stricter defaults for under-18s, the Sensitive Content Control, and the supervision tools in Family Center.
• Snapchat Family Center, together with the content controls that limit what is surfaced in Stories and the Spotlight feed.

If a feed has already drifted, a focused twenty-minute session can reset the slate — and it works best done together, as maintenance rather than punishment:

1. Open the main apps side by side with your teen.
2. Review the following and subscription lists; mute or unfollow accounts that leave them feeling worse.
3. Clear or pause watch and search history wherever the platform allows it.
4. Mark ten or fifteen unwanted posts as "not interested," giving the recommender an explicit signal.
5. Switch on teen, restricted, or sensitive-content modes.
6. Have your teen follow or search a few interests they genuinely value, to seed the feed with healthier signals.
7. Check back on their mood, sleep, and the tone of the feed after a week.

Settings, though, are the smaller half. The larger half is the ongoing conversation — and the most effective version of it is not a lecture about danger but a genuine, recurring curiosity about what your teen is actually seeing. Ask them to show you what tends to come up. Ask what is funny, what is boring, what is upsetting. A teenager who can talk to you about their feed without being judged keeps a line open for the day something on it frightens them — and that open line is worth more than any blocked app.

When you do open the conversation, aim for curiosity rather than interrogation — questions that invite a teen to show you their world, not to account for it. A few openers that tend to land:

• "Can you show me what your feed has been giving you lately? I’m curious how different it is from mine."
• "Does this app ever leave you feeling worse once you’ve closed it?"
• "Are there topics that keep coming back even when you don’t want them?"
• "Let’s tune the feed together — this isn’t about taking the phone away."

Each one treats your teen as the expert on their own feed, which is both true and disarming.

Two things make that conversation hold. The first is to teach the machinery rather than only the rules. A teenager who understands why their feed leans the way it does — that strong emotion is being optimized for, that lingering is read as a vote, that the system has no idea whether a clip helped or hurt — gains a kind of immunity that no blocklist provides. They begin to notice the feed working on them, and noticing is most of the defense. The second is to model it. A household where the adults also put phones down at dinner, also talk about the clip that annoyed them, and also admit to losing an hour to a feed is teaching by example; a rule that applies only to the teenager reads as control, and control is the thing this guide keeps warning sits between you and the visibility you need.

Some parents will also decide that, after a genuine concern, they want more direct visibility for a while. In many places a parent or legal guardian may use age-appropriate monitoring on a minor’s device — the rules vary by country, state, and custody situation, so check what applies where you live — and how you do it matters more than whether you do it. Covert surveillance, if a teen discovers it, confirms that adults cannot be trusted and teaches them to route around you onto a device you cannot see. Transparent, age-appropriate monitoring — your teen knows the tool is there, knows what it does, and knows it exists because something genuinely serious happened — works with the relationship instead of against it. Think of it as scaffolding: visible, temporary, and taken down deliberately as your teen earns back the autonomy it was protecting.

Reporting and resources

When your teen encounters genuinely harmful content, reporting it is worth the few minutes it takes. Use the in-app report tool on every platform — it is the fastest route to removal and it trains the platform’s own systems. Content that sexually exploits a minor is different and more serious: report it in the United States to the NCMEC CyberTipline, and in the United Kingdom to the Internet Watch Foundation. If your teen is in crisis, contact the 988 Suicide & Crisis Lifeline in the US or Childline in the UK; elsewhere, your national crisis line. For background and current platform guidance, the bodies below publish free, regularly updated material.

• For the evidence — the U.S. Surgeon General’s Advisory on social media and youth mental health, and the APA Health Advisory on adolescent social media use.
• For platform setup — Internet Matters and Common Sense Media, both of which keep current, platform-by-platform parental-control and content-setting guides.
• For how teens really use platforms — the Pew Research Center’s ongoing work on teens, social media, and technology.
• For crisis support — the 988 Suicide & Crisis Lifeline in the US and Childline in the UK.

Frequently asked questions