Australia spent last year pulling under 16s off social media. The minimum age law has been in force since 10 December, with fines up to $49.5 million for platforms that fail to take reasonable steps. OpenAI has now gone the other way and built a product aimed squarely at 13 to 17 year olds.
ChatGPT for Teens is a separate experience with its own safeguards, its own model rules and parental controls attached. For anyone running a business the interesting part is not the study features. It is that OpenAI decides who lands there by estimating the user’s age.
What ChatGPT for Teens actually is
The product bundles Study Mode, quizzes, learning visualisations and a new Study Hours setting that turns guided study on at chosen times, according to OpenAI. There are homework reminders that spot when a teen is trying to shortcut an assignment and push them back towards step by step work.
Parents with a linked account get Quiet Hours, control over selected settings and safety notifications in high risk situations, now extended to eating disorders. OpenAI also announced a partnership with CodeAI aimed at teachers and classrooms.
The age is estimated, not asked
This is the line that matters. A user lands in the teen experience if they state an age between 13 and 17, or if OpenAI’s system estimates they are under 18. Age prediction has been rolling out for months, and this is what it feeds.
Estimation runs on everyone. That is the quiet implication: to work out who is a teenager, the system has to form a view about every account, including the ones your staff log into at work.
What the model may no longer do
The under 18 model spec goes past blocking romantic roleplay. ChatGPT should not use romantic language, should not encourage emotional dependence, and should not imply it has feelings or consciousness. Break reminders and product cues repeat that it is AI.
OpenAI has also published under 18 evaluations in its system cards, covering self harm, eating disorders, violence, age restricted goods and sexual content. Those are new, and they are the first numbers anyone outside the company can point at.
Why Australian businesses should pay attention
Three reasons, and none of them are about parenting. The first is the eSafety Commissioner, who has already issued transparency notices to AI companion apps and published the findings. The prohibitions OpenAI just wrote into its spec sit close to what that work was examining.
The second is age assurance as a compliance category. Australia now has a working regime for it in social media and the technology trial behind it is public. A vendor that estimates user age here operates where data residency already became a purchasing question.
The third is practical. If age estimation misfires on an adult, that person gets a restricted product and no explanation. For anyone standardising on ChatGPT across a team, that sits in the same AI governance conversation as agent access.

The question nobody has answered
The minimum age law exempts services whose main purpose is education, among other categories. OpenAI has titled this launch around learning. Whether that framing has any bearing on how the law reads a general purpose assistant is not something either side has addressed.
There is also a gap in the announcement itself. OpenAI names no launch date and no country list, so whether Australian teenagers can use this today is unconfirmed. The company has form here: Computer History reached Australia on day one last week and left Europe out.
For Australian operators the useful posture is to assume it arrives and settle the policy first: who in the business uses a consumer AI account at all, and what happens when the vendor starts sorting users into tiers nobody chose.
