How the recommendation algorithms behind apps like TikTok, YouTube, and Instagram can push children toward extreme, distressing, or addictive content — and how to reset and manage them.
Algorithmic Amplification is the way social and video apps' recommendation systems push increasingly engaging, extreme, or distressing content to keep a child watching, creating rabbit holes and filter bubbles.
Most apps children use decide what to show next with a recommendation algorithm designed to maximise engagement — how long you watch and how much you interact. To do that, algorithms often amplify content that is sensational, extreme, or emotionally charged. For a child, a brief moment of curiosity can turn into a feed dominated by one topic — a 'rabbit hole' — whether that is harmless (a hobby) or harmful (disordered eating, misogyny, self-harm, or conspiracy content). Recommender systems and the harms they can amplify are a specific focus of the UK Online Safety Act and Ofcom's codes of practice.
Every pause, rewatch, like, follow, and search teaches the algorithm what keeps a user hooked, and it serves more of the same — plus more extreme versions to sustain attention. Because the system optimises for engagement rather than wellbeing, distressing or divisive content often travels furthest. Children are especially affected: they may not realise the feed is curated rather than neutral, and infinite scroll plus autoplay remove the natural stopping points. The result can be narrowed worldviews (filter bubbles), pressure and comparison, and exposure to progressively harmful material without ever searching for it.
In your child's behaviour
On their device
Teach how the feed is built
Explain that the app chooses what to show to keep them watching, that it is not neutral or 'the truth', and that engaging with something — even out of shock — brings more of it. Understanding the mechanism is the strongest defence.
Actively train and reset the algorithm
Show your child how to use 'not interested', report, unfollow, and clear/reset recommendations, and to deliberately engage with positive content. On several apps you can refresh the For You feed to break a rabbit hole.
Design in stopping points
Turn on wellbeing and take-a-break reminders, disable autoplay where possible, use screen-time limits, and keep devices out of bedrooms overnight so infinite scroll doesn't replace sleep.
In immediate danger: call 999. For non-emergency police matters, call 101.
Concerned about a child but it's not an emergency? NSPCC helpline 0808 800 5000. Childline for young people 0800 1111.
This is practical educational content to support families. For case-specific concerns about a child's safety, contact the NSPCC helpline on 0808 800 5000 or your local safeguarding team.
Myth: The algorithm just shows my child what they've searched for or chosen to watch, so it's basically neutral.
Fact: Recommender feeds are built to maximise watch time, not to be neutral or accurate. Ofcom's own research into recommender systems found that personalised 'For You' feeds are children's main pathway to encountering harmful content, often material they never searched for, which is why the Online Safety Act's children's safety codes require higher-risk services to configure these algorithms to filter the most harmful material out of children's feeds.
Myth: If something upsetting comes up, my child can just scroll past and ignore it, so no real harm is done.
Fact: Ofcom's research shows recommender systems respond rapidly to what a child views and interacts with, so pausing on or rewatching distressing content, even out of shock, signals the algorithm to serve more of it. Scrolling past is not enough on its own; actively using 'not interested', reporting, unfollowing and resetting recommendations is what actually shifts the feed.
Myth: As long as my child's account is private or they only follow friends, recommended content can't reach them.
Fact: Account privacy controls who can see a child's own posts, not what gets recommended to them. Ofcom's research into recommender systems found that 'For You' style feeds are shaped by a user's behaviour and interactions across the whole platform rather than by their follow list, so a private account does not stop unrelated or harmful material being surfaced.
At this age, most viewing still happens in supervised or semi-supervised spaces such as YouTube Kids or shared-device sessions, so lean on parental controls and kids' modes rather than relying on the general algorithm to behave itself. Sit with your child during longer viewing sessions and notice whether one topic or creator is starting to dominate; if it is, use the supervised account's controls to reset recommendations together. Keep autoplay off where the app allows it and treat video time as something you both end at an agreed point, not something the feed decides for you.
This is often when children start using apps built around a 'For You' style feed, so talk them through how it decides what to show before problems arise rather than after. Set up any family-pairing or supervision tools the app offers, agree together which content is off-limits, and show them exactly how to tap 'not interested', report a post, and reset their recommendations. Check in on what their feed actually looks like from time to time, since at this age they may not yet recognise or flag a rabbit hole themselves.
Teenagers at this age usually manage their own feeds and are more likely to be pulled into rabbit holes around serious themes such as self-harm, extreme body ideals, or divisive worldviews, often without meaning to search for them. Keep the conversation going about how engaging with something, even out of shock, trains the algorithm to serve more of it, and encourage them to actively curate their feed rather than passively consume it. Watch for warning signs such as a feed fixated on one distressing theme or a mood that tracks with time spent scrolling, and raise it calmly rather than confiscating the device.
By this age your role shifts from managing settings to reinforcing judgement, since your young person will increasingly use these platforms unsupervised and make their own choices about what to engage with. Talk about how algorithmic feeds can narrow a person's worldview over time, particularly around news, politics, or body image, and encourage habits such as periodically resetting recommendations and deliberately diversifying who they follow. Treat this as a lifelong media-literacy skill they are building, not a rule you are still enforcing.
Was this page helpful?
Get practical child safety updates in your inbox. No spam.