Algorithms increasingly shape what children watch, read, buy, and discuss online. Because these systems are designed to predict attention and personalise content at scale, the question is no longer whether children encounter algorithms, but whether digital services are sufficiently designed around children’s safety, privacy, and developmental needs. The strongest case is not for excluding young people from digital spaces altogether, but for placing meaningful safeguards around the systems that structure those spaces.
The Invisible Editor of Childhood
Recommendation systems decide what appears next in a video feed, search result, social media timeline, or online game. They often rely on behavioural signals, such as viewing time, clicks, shares, and repeated searches, to infer a user’s interests. For children, whose preferences and identities are still developing, this can create a particularly influential environment. What seems like a neutral sequence of suggestions can gradually shape attention, social comparison, consumer habits, and perceptions of what is normal.
The scale of this issue is substantial. The United States Surgeon General reported that up to 95 percent of young people aged 13 to 17 use at least one social media platform, while more than one third say they use social media almost constantly. The advisory also noted that nearly 40 percent of children aged 8 to 12 use social media despite many platforms setting a minimum age of 13. These figures do not establish that social media is inherently harmful, but they show why the design of recommendation systems deserves serious scrutiny.
Why Engagement is Not a Child Safety Standard
Many algorithms are optimised to increase engagement, meaning they are intended to keep users watching, scrolling, clicking, or returning. This commercial objective may conflict with children’s interests when emotionally intense, sensational, or highly repetitive material captures attention more effectively than balanced content. A child who pauses on one distressing video, for example, may be shown more material with similar themes, even when that pattern is not beneficial to their wellbeing.
Ofcom, the United Kingdom’s communications regulator, has concluded that personalized recommendations are children’s principal route to encountering harmful content online. This shifts the discussion beyond individual parental supervision. Even attentive families may struggle to monitor systems that refresh content continuously and make decisions through processes that users cannot fully see or control.
The American Psychological Association similarly argues that online functionality should reflect young people’s developmental capacities rather than simply reproduce services designed for adults. Its guidance highlights recommended content as a feature that may be especially difficult for younger users to resist, and stresses that youth safety and welfare should take priority over engagement.
Protection Without Digital Exclusion
Protecting children from harmful algorithmic effects does not require treating all technology as harmful. Digital platforms can support education, creativity, community, and access to information. The policy challenge is therefore to distinguish between participation and exposure to persuasive design.
Practical safeguards could include private accounts by default, clearer explanations of why content is recommended, options to reset personalised feeds, and limits on recommendations based on extensive behavioural profiling. Platforms could also give greater weight to choices that children deliberately express, rather than relying mainly on inferred interests drawn from continuous monitoring.
This approach is reflected in recent European regulatory guidance. The European Commission’s guidelines on protecting minors online recommend privacy by default and safer recommender systems, including measures to reduce exposure to harmful content and avoid excessive monitoring of minors’ online behaviour. The Digital Services Act also prohibits advertising based on profiling when a platform knows with reasonable certainty that its recipient is a minor.
Shared Responsibility Matters
Parents and schools remain important in helping children develop media literacy and healthy digital habits. However, individual responsibility alone cannot address platform designs that are complex, opaque, and deliberately optimised for sustained attention. Companies that build and profit from these systems should bear a corresponding responsibility to assess foreseeable risks, adapt features for younger users, and allow independent oversight.
At the same time, safeguards should respect children’s privacy, evolving autonomy, and access to beneficial information. Overly intrusive age checks or blanket restrictions may introduce new privacy concerns and disproportionately limit young people’s participation. Effective protection should therefore be proportionate, evidence based, and focused on reducing high risk design choices rather than simply restricting access.
A Final Note
Children do not need an algorithm-free childhood, but they do need digital environments that do not treat their attention, data, and vulnerability as resources to be exploited. Protecting them means redesigning online systems so that safety and development are built into the experience from the beginning.
