The playground we would never allow

A thousand children use a community playground each week. We would not tolerate two hundred being harmed, nor twenty, nor even one child harmed reliably every single week, especially if the operator knew and did nothing. The same standard should apply to social media and AI chat products used by children. The piece closes with real Australian findings from the eSafety Commissioner and from peer-reviewed research.

0 children, every week

The playground

A thousand children. Every week.

Picture a playground here in our community. A thousand kids play on it, week after week. We want it busy. We want it loved.

200 of 1000

Now imagine two hundred came home harmed.

Not once. Every week. Nobody would call that bad luck. The gates would be locked by Friday and someone would answer for it.

20 of 1000

What if it were only twenty?

Still a scandal. Still unthinkable. The number has shrunk, but the answer has not changed.

1 of 1000

What if it were just one?

Here most of us pause. Playgrounds carry some risk, and we accept that a child can have an accident. That is part of life, and we know it.

One. Every week.

But what if it were one child, every week.

The same playground. The same swing. Predictable as the calendar. Would we still call that an accident? Or would we start asking why nobody had fixed it?

And they knew

What if the operator had the report on their desk.

Telling them it would keep happening. And they left the playground exactly as it was.

That is the part we would not stand for.

The same harm, a different space

This is roughly what we are living with.

On social media. On AI chat products. The harm is known. It recurs. And too often it has not been addressed as it should be.

Where the line sits

We are not asking for a world with zero risk.

We are saying there is a kind of harm we will not have built into the spaces our children use. And that those who own these products carry the duty to make them fit for the people, especially the young people, who actually use them.

From the record

This is not hypothetical.

These are insights from Australian research and lived experience. The parable above uses simple, made-up numbers to make a point; the findings below are real.

The playground is a story. These are not. In Australia, the bodies whose job is to examine these harms have already done so, and they have set out what they found.

eSafety Commissioner · March 2026

AI companion apps, and the children using them

Australia's eSafety Commissioner examined four popular AI companion services, Character.AI, Nomi, Chai and Chub AI, after issuing them legal notices.¹ It found "serious gaps in basic safeguards for children".² None had robust age checks; most relied on a tick-box at sign-up. Most did not reliably point users to support when self-harm or suicide came up in chat. And several did too little to stop their own tools producing sexual material involving children.

79% of Australian children aged 10 to 17 told the eSafety Commissioner they had used an AI companion or assistant. Around 200,000 had used an AI companion specifically.¹ (From a survey of 1,950 children.)

Peer-reviewed research · 2026

Harmful content online is common, and young people see more of it

A national study led by Orygen (Australia's national youth mental health research organisation) and the University of Melbourne, published in BMC Public Health, surveyed about 3,500 Australians.³ Just over half had come across self-harm or suicide-related content on social media. People aged 15 to 25 were roughly four times more likely to encounter it than older adults. For most, seeing it worsened their mood, and a smaller group reported harming themselves after seeing it, again more often among the young. We include it here because the same dynamic carries straight over to AI chat products, where this content can surface inside conversations built to feel personal, and where the line between AI assistants and companions is already blurring.

What you can do

Have the awkward conversation. The eSafety Commissioner has plain guides for parents, carers and young people at esafety.gov.au.
Know that a self-declared age at sign-up is not a real safeguard, and that Australia's age-restricted material codes now apply to companion chatbots.
Bring it to the Commons. This is exactly the kind of local line we are here to draw together.

If any of this is close to home

You can talk to someone now. Lifeline 13 11 14. Kids Helpline 1800 55 1800 (for ages 5 to 25). Beyond Blue 1300 22 4636. In an emergency, call 000.

Sources

eSafety Commissioner, "eSafety report shows AI companions are putting children at risk" (24 March 2026). esafety.gov.au
eSafety Commissioner, "Findings from transparency notices on AI companion apps" (October 2025). esafety.gov.au
Robinson, J., La Sala, L., Kenny, B. et al. How do Australian social media users experience self-harm and suicide-related content? A National cross-sectional survey comparing young people and adults. BMC Public Health 26, 384 (2026). https://doi.org/10.1186/s12889-025-25646-0

Help us get this right

This resource is a work in progress. If something here does not match what you, your family, your school or your workplace are seeing, or if you can help us sharpen it, we would like to hear from you.

The playground we would never allow.