AI companion apps are widely used by teenagers and much less understood by schools than general AI assistants. The safeguarding considerations are genuinely different — here's what DSLs and pastoral teams should know.
What these apps are
Companion chatbots are designed to simulate a relationship — a friend, a confidant, sometimes a romantic partner. Unlike a general assistant answering questions, they're built for ongoing emotional engagement, remember previous conversations, and are deliberately designed to keep users coming back. Many are free, most are trivially easy to access, and age checks are often minimal.
Why the risks differ from general AI
- Emotional dependency — the app is engineered to feel like a relationship, and lonely or struggling young people are the most susceptible
- Displacement of real support — a teenager confiding in a bot may not confide in an adult who could actually help
- Inappropriate content — some are explicitly designed for romantic or sexual roleplay, with weak age verification
- Harmful responses in crisis — these are not clinical tools and may respond badly to disclosures of self-harm or abuse
- Data — young people disclose deeply personal information to a commercial product
- Validation without challenge — a companion designed to please won't push back on distorted thinking
The safeguarding concern that matters most: a young person in distress who is confiding in a chatbot is a young person not confiding in someone who could get them help. That displacement is the risk to watch for, more than any single piece of content.
Signs worth noticing
Nothing here is diagnostic on its own, but pastoral teams might note: heavy secretive phone use with an app staff don't recognise; a young person referring to an online friend nobody has met; withdrawal from real friendships alongside it; or distress when access is removed. As always, curiosity rather than accusation is the right opening.
How to respond
- Don't ridicule it — the attachment feels real, and mockery ends the conversation
- Be curious about what the young person gets from it — usually it points to an unmet need
- Follow normal safeguarding routes if there's any disclosure or risk indicator
- Address the underlying need — loneliness, anxiety, lack of a trusted adult
- Involve parents appropriately, recognising many won't know these apps exist
What schools should do proactively
Cover it in RSHE and online safety — how these apps are designed to retain users, why an AI cannot actually care, and why real relationships matter. Brief pastoral staff so they recognise it. And tell parents, because most have no idea this category exists. See what parents should know about AI and AI image risks for DSLs.
Keep it proportionate
Most young people who try these apps come to no harm, and framing every user as at risk would be wrong. The concern is specific: vulnerable, isolated young people forming a dependency that replaces human support. That's who your attention belongs on.
Want pastoral staff briefed on this?
We run sessions covering the AI risks schools aren't yet talking about. AskColin gives schools low-cost monthly support with one-to-one help whenever you need it — so when you're stuck on a real job at 8am, there's someone to ask. Practical, jargon-free, built around your team.
Get one-to-one AI support for your schoolFrequently asked questions
What are AI companion chatbots?
Apps designed to simulate a relationship — a friend, confidant or romantic partner. Unlike general assistants, they're built for ongoing emotional engagement, remember previous conversations and are designed to keep users returning. Many are free with minimal age checks.
What are the safeguarding risks of AI companions for teenagers?
Emotional dependency by design, displacement of real support so a struggling young person confides in a bot rather than an adult who could help, inappropriate sexual content with weak age verification, poor responses to crisis disclosures, data disclosure to commercial products, and validation without healthy challenge.
What signs might suggest a pupil is using AI companion apps?
Heavy secretive phone use with an unfamiliar app, referring to an online friend nobody has met, withdrawal from real friendships, or distress when access is removed. None is diagnostic alone, and curiosity rather than accusation is the right approach.
How should schools respond to AI companion use?
Don't ridicule it — the attachment feels real. Be curious about what the young person gets from it, since it usually points to an unmet need. Follow normal safeguarding routes for any disclosure, address the underlying need, and involve parents who often don't know these apps exist.