OpenAI is making the case that teenagers should have access to AI, just not the same AI adults get. In a new post, the company lays out how it’s building ChatGPT for younger users: age-appropriate protections, learning tools, parental controls, and partnerships with outside experts. The framing matters. OpenAI isn’t arguing teens should be locked out. It’s arguing they should be let in through a different door.
Here’s what that means in practice, and how to actually set it up.
Quick Start
What you’ll learn: how OpenAI’s teen safety approach works, and the steps to configure ChatGPT for a teenager in your house.
What you need: a ChatGPT account for yourself, a ChatGPT account for your teen, and about ten minutes.
Who this is for: parents, teachers, or anyone handing an AI tool to someone under 18 for the first time. No technical background needed.
Step 1: Understand what “age-appropriate” actually means
Before you touch a setting, know the model. OpenAI says ChatGPT applies different protections depending on whether the user is a teen or an adult. Same product, different guardrails.
Why this matters: a lot of parents assume AI safety is one big on/off switch. It isn’t. The system tries to figure out who it’s talking to, then adjusts. That means the accuracy of your teen’s account details actually affects the experience they get.
Step 2: Set up the teen’s own account properly
Don’t let a 15-year-old use your account. It’s tempting, it’s faster, and it defeats the entire system.
Why this matters: protections keyed to age can’t work if the account says the user is a 43-year-old. Every safeguard OpenAI describes depends on the platform knowing it’s talking to a teen. A shared login makes your teen invisible to the exact protections built for them.
Step 3: Turn on parental controls
OpenAI lists parental controls as a core part of the teen experience. Go into your account settings, link your teen’s account, and walk through the options.
Why this matters: controls that exist and controls that are switched on are different things. Most safety features ship off by default, and defaults are where good intentions go to die. Do it the same day you set up the account, not “this weekend.”
A tip while you’re in there: read each option with your teen in the room. It turns a surveillance move into a conversation.
Step 4: Point them at the learning tools
OpenAI highlights learning tools as part of why teens deserve access at all. This is the part most safety guides skip.
Why this matters: if the only thing you do is restrict, you’ve taught your teen that AI is contraband. Teens who understand AI as a study aid, a tutor, a thing that explains a concept five different ways until one lands, use it very differently from teens who use it to hide homework. Show them the good use case first. The restrictions land better after.
Step 5: Revisit it
OpenAI says it works with outside experts on these protections, which is a polite way of saying the rules will keep changing. They should. Nobody has this figured out yet.
Why this matters: a setting you configured in March may not reflect what the product does in September. Put a recurring reminder in your calendar. Twice a year is enough.
Why this is significant
What stands out is the position OpenAI is taking. The easy move for a company under regulatory pressure is to raise the age gate and walk away. Instead OpenAI is arguing that teens will use AI regardless, so the responsible path is building a version that’s safe for them rather than pretending they aren’t there.
That’s a harder promise to keep. Age prediction is imperfect. Teens are famously good at routing around restrictions. And “age-appropriate” is a judgment call that no model makes perfectly.
But the alternative, where teens use adult-tuned AI through borrowed logins with zero protections, is worse. Much worse.
Next steps
- Check whether your teen’s school has an AI policy. It probably conflicts with something.
- Ask your teen what they actually use it for. The answer is usually more interesting than you expect.
- Watch for updates. This category is moving fast, and today’s settings page won’t be next year’s.
Full details are in OpenAI’s original post.