Education news

One Question Tells You If Your Teen’s AI Tutor Teaches

OpenAI has given teens a ChatGPT that tutors instead of answers, and its own education chief says that difference decides whether it helps or hurts. You do not have to trust the marketing. One question tells you which kind of tool your child is using.

Laura Lurns Last updated August 23, 2026 · 10 min read

One Question Tells You If Your Teen’s AI Tutor Teaches

If your teenager keeps ChatGPT open while doing homework, you have probably wondered whether it is teaching them anything or quietly doing the assignment for them. On August 18, 2026, OpenAI launched a version of ChatGPT built for teens, and it turns on a feature called Study Mode by default, one designed to ask guiding questions instead of handing over answers. The company’s own head of education draws exactly the line a parent cares about: prompted to tutor, the tool helps; used as an answer machine, it hurts. That distinction is the whole ballgame, and it is not something you have to take OpenAI’s word for. There is one question worth asking of this tool, or any AI study app your child uses, that tells you which kind you are looking at.

TL;DR

OpenAI launched ChatGPT for Teens on August 18, 2026, auto-enrolling users aged 13 to 17 (or estimated under 18) into a teen experience with Study Mode switched on by default.
Study Mode, first released in July 2025, withholds direct answers and instead uses guiding questions, scaffolding, and knowledge checks, rather than acting as an answer engine.
OpenAI’s Head of Youth and Families said Study Mode has been tested in randomized controlled trials showing “positive gains,” but declined to provide figures, and no independent study has been published.
Study Mode runs on the general ChatGPT model with custom instructions, not a purpose-built tutoring system trained on vetted academic content, per MIT Technology Review.
Learning research points to effortful retrieval, step-level interaction, and next-step feedback as what makes study tools work, though none of those studies tested Study Mode itself.

Context the coverage leaves out

What the coverage gets wrong

The launch got covered two ways, and both miss the same beat. In one telling, AI is about to tutor every teenager to success; in the other, it is a new way for kids to cheat. Both argue about the screen and skip the mechanism, which is the only thing that decides whether a study tool teaches. A second problem: OpenAI’s claim that Study Mode passed randomized trials with “positive gains” keeps getting repeated as if it settles the matter, when the company has published no figures and no independent study exists. It is a claim to note, not evidence to lean on. And watch for coverage that describes the scheduling feature, Study Hours, in the language of memory science. It decides when the mode turns on. It is not spaced practice, and calling it that lends the product a rigor it has not shown.

OpenAI turned on a tutor-style Study Mode for teens by default. Here are the questions parents are asking about what it means at home.

Common questions

Is ChatGPT’s Study Mode actually good for my teen’s learning?

The honest answer is that no one outside OpenAI has proven it yet. OpenAI says Study Mode was tested in randomized controlled trials with positive gains, but it has published no figures and no independent study exists. What the design claims to do, ask guiding questions and check understanding instead of handing over answers, lines up with how learning research says memory forms. Whether the tool as built delivers that consistently is the open question. Treat it as promising and unproven, and watch how your teen actually uses it.

How do I tell if an AI study tool is teaching or giving answers?

Watch one session. A tool that teaches makes your child do the thinking: it asks questions, waits for an answer, and nudges when the reasoning goes sideways. An answer machine produces the solution with a tidy explanation your child scrolls past. The test is portable, working on Study Mode, on a competitor, or on a free chatbot. The label on the product matters less than what it makes your child do.

My younger child wants to study with AI too. Is it the same?

The mechanism test is the same, but the age question is not. This product is built for ages 13 to 17. For younger children, screen-delivered practice is a weaker fit as a main way to learn, no matter how well the tool is designed, because young children build skills through hands-on practice and interaction with a person. Use the same question about producing the answer, and add a firmer limit on how much of the learning happens on a screen at all.

My teen is struggling in school even with study tools. Should I get them evaluated?

A screener or an evaluation helps you understand where your child is struggling and where to start, and starting sooner beats waiting to see. A screener is a starting point, not a diagnosis. If your child might need formal accommodations such as an IEP or 504 plan, or you suspect a vision, hearing, or medical cause, pursue a professional evaluation too, because that is the only route to those supports. Whatever a tool or a report says, the most reliable help is steady, direct practice on the specific skill that is hard, alongside an adult who knows your child.

Share the idea

OpenAI’s teen ChatGPT defaults to a tutor mode that asks instead of answers. The test for any AI study tool: does it make your kid produce the answer, or hand it over?

Post on X Follow

What OpenAI actually launched

On August 18, 2026, OpenAI began rolling out ChatGPT for Teens, a version of its chatbot built for users aged 13 to 17, with a global rollout the company expected to finish within about two weeks. Teens who identify themselves as under 18, or whom an age-prediction system estimates to be under 18 based on how and when they use the app, are enrolled automatically. Parents get controls over quiet hours, content sensitivity, and safety notifications. And a feature called Study Mode, which had been optional for all ChatGPT users since July 2025, is now switched on by default for teens.

Study Mode is the part that matters for homework. Instead of spitting out an answer, it is designed to work like a Socratic tutor: guiding questions, step-by-step scaffolding, prompts that ask the student to explain their thinking, and knowledge checks along the way. OpenAI says the instructions behind it were developed with pedagogy experts from more than 40 institutions. A companion setting, Study Hours, lets a parent or teen schedule times when Study Mode is on by default, and a Responsible Homework Reminder is meant to notice when a teen is trying to shortcut an assignment and steer them back toward working the problem.

Does it work? OpenAI says yes, but has not shown its work. Its Head of Youth and Families, Lauren Jonas, said the company has had Study Mode “in the field in a number of randomized controlled trials and are seeing positive gains in performance,” then declined to provide any figures. No study has been published, so there is no sample size, no comparison group, and no result for a parent to weigh. Worth knowing too: Study Mode is not a separate tutoring system trained on vetted academic material. It is the standard ChatGPT model with a set of custom instructions layered on top, which means it still generates the occasional confident, wrong answer, as MIT Technology Review noted.

When ChatGPT is prompted to teach or tutor, it can significantly improve academic performance. But when it’s just used as an answer machine, it can hinder learning.
Laura Lurns · Learning Success expert

What the learning science actually says

Coverage of AI in schools tends to split into two camps: the tool that will personalize education for every child, and the tool that will let a generation cheat its way through school. Both put the screen itself on trial and skip the question that actually decides the outcome, which is what the tool makes a child do. OpenAI’s own VP of Education, Leah Belsky, put it plainly when Study Mode first launched: “When ChatGPT is prompted to teach or tutor, it can significantly improve academic performance. But when it’s just used as an answer machine, it can hinder learning.” That is the company naming the right standard for itself.

Learning research has been pointing at that same line for a long time. Being made to produce an answer from memory, rather than being shown it, is what makes learning stick (Roediger and Karpicke). Software that engages a student at the level of each step, catching where the reasoning breaks down and prompting the next move, tends to help more than software that only marks the final answer right or wrong (VanLehn; Kulik and Fletcher). And feedback that tells a learner what to do next does more than a bare score (Kluger and DeNisi; Wisniewski, Zierer and Hattie). A tool whose loop is ask, make the student answer, then check is built on something real.

Here is the catch, and it is the honest part. That research validates a category of design, not this particular product. No one has independently tested whether Study Mode’s actual behavior delivers genuine step-by-step teaching, or whether a general-purpose chatbot is performing a convincing imitation of it. And the deeper problem it is swimming against is that most study apps quietly swap a real learning mechanism for a cheap engagement signal: points, badges, streaks, minutes logged. It is entirely possible for a child to build an impressive streak and a long record of screen time while the learning underneath barely moves. Engagement is not the same thing as learning. What makes Study Mode an interesting test case is that it says so out loud. The open question is whether the feature as built clears the standard its own maker set, or whether naming the standard is the marketing.

Key takeaways

Study Mode is on by default: OpenAI’s teen ChatGPT guides with questions instead of handing over answers.
The portable test: Ask whether a tool makes your child produce the answer or hand it over.
Engagement is not learning: Retrieval and next-step feedback build skills; points and streaks mostly hold attention.

The one question worth asking

You do not have to adjudicate OpenAI’s trials, and you do not have to become an expert in artificial intelligence. Watch your teen use the thing for ten minutes and ask one question: is it making them produce the answer, or handing it over? A tool that teaches asks, waits, and nudges when the thinking goes off track. An answer machine hands back a finished solution wrapped in an explanation your child scrolls past on the way to the next problem. The same question works on Study Mode, on a rival app, on a plain chatbot with no study features at all.

This is not a verdict for or against letting your teen use AI. It is a way to tell a good use from a hollow one. Two things are worth holding onto while you decide. Even the best-designed tool is a supplement, not a substitute, for a teaching adult and the relationship around the learning, and no chatbot changes that. And be careful not to over-read the scheduling features: Study Hours decides when Study Mode switches on, which is useful for building a routine, but it is a timer, not a memory-science feature that decides what your child practices or when they see it again.

One more thing, because this test will travel further than the product it came from. ChatGPT for Teens is built for ages 13 to 17. If you have a younger child who wants to study with AI, the mechanism question still holds, but add a firmer one about the screen itself, because for young children screen-delivered practice is a weak fit as a main way to learn no matter how clever the design. For a teenager, the tool is here and it is on by default. The most useful thing you own is not a setting. It is the question that tells you, in ten minutes of watching, whether your child is doing the learning or watching it happen.

You do not need to understand how a language model works to tell whether one is helping your child. You need the right question, and it is the same one that separates real learning from its imitation: is my child producing the answer, or being handed it? The villain here was never AI, and it was never the screen. It is engagement mistaken for learning, the polished loop that keeps a child busy while the skill underneath stays exactly where it was. That is the trap Learning Success was built to avoid. Our All-Access membership opens an assessment that asks about every processing system your child’s learning draws on, and a roadmap that names what to build first. It is a starting point for helping your child, not a diagnosis. If your child might need formal accommodations such as an IEP or 504 plan, or you suspect a vision, hearing, or medical cause, pursue a professional evaluation too, because that is the only route to those supports.

See what All Access gives your child

Is your child struggling in school?

Get your free personalized learning roadmap

You describe what you see at home. We turn it into a plan you start this week.

Answer 5 short questionnaires about what you already notice, 30–45 minutes at your own kitchen table
Your child sits no test and gets no score: nothing to schedule, nothing for them to dread
You do the answering, the AI does the writing, and a person reviews it before it reaches you
Access all 40+ courses instantly: reading, math, focus, processing and more, with new ones added regularly

Why we use AI, plainly: it writes from a knowledge base our team maintains and audits. We work through it line by line and pull anything the evidence stops supporting. The roadmap you get on Tuesday reflects what we corrected on Monday, and a human still reads it before you do.

Your school district must evaluate your child free of charge if you ask in writing, whatever your income and whatever the outcome (US, 34 CFR 300.111 and 300.301(b)). That route takes time and answers a different question than you do. This one starts today, from what you already know.

Get my free assessment & 7-day trial Cancel anytime during your trial.

Your answers stay yours. We do not sell your personal information, and we do not hand identifiable assessment data to outside AI companies to train their models.

A screener is a starting point, not a diagnosis. If your child might need formal accommodations (an IEP or 504 plan), or you suspect a vision, hearing or medical cause, pursue a professional evaluation too. That is the only route to those supports.

References

OpenAI — ChatGPT for Teens (August 18, 2026)
OpenAI — Introducing Study Mode (July 29, 2025)
TechCrunch — OpenAI launches a safer ChatGPT for teens
The Next Web — ChatGPT for Teens: OpenAI’s age prediction and Study Mode
VentureBeat — OpenAI’s Study Mode helps students learn step by step
MIT Technology Review — OpenAI is launching a version of ChatGPT for students

Laura Lurns · Learning Success expert Writes about the learning brain for parents who want plain answers. Every article is grounded in current neuroscience and classroom practice.