On October 7, 2026, OpenAI published its first progress report on ChatGPT for Teens. The headline claim: teens on the experience spend under 15 minutes a day on ChatGPT on average. Alongside the data, OpenAI announced College Planner (coming soon), flashcards, easier quizzes from uploaded notes, and a three-year teen advisory effort with Boston Children's Hospital's Digital Wellness Lab.
The timing matters. The same day, Common Sense Media's Youth AI Safety Institute rated the product an "Unacceptable Risk" in a separate test, which we cover in our write-up of that assessment. This post sticks to what OpenAI said, what the numbers can and cannot tell you, and what the new tools do.
TL;DR: what OpenAI announced
| Item | What OpenAI says | Status |
|---|---|---|
| Average daily use | Under 15 minutes a day for teens | Reported data |
| Long sessions | Fewer than 2% of teens exceed three consecutive hours | Reported data |
| Break reminders | In almost half of reminded conversations, teens took a break or ended the chat within five minutes | Reported data |
| Learning Visualizations | Nearly 1.2 million teens used them in one week | Reported data |
| Study Mode | More than 180,000 teens used it in one week | Reported data |
| College Planner | One plan for requirements, deadlines, tasks and aid steps; US grades 10 to 12 first | "Soon" |
| Flashcards and quizzes | Build decks and quizzes from notes; decks saved in the Library | Announced |
| Multi-photo capture | Merge note photos into one PDF | iOS now, Android in progress |
| Teen council | Digital Wellness Lab Student Advisory Council, three years | Funded |
What ChatGPT for Teens is, briefly
ChatGPT for Teens is the default experience applied to accounts identified as under 18. OpenAI launched it on August 18, 2026 with stronger built-in safety protections, parental controls, and learning tools such as Study Mode. Our launch explainer walks through the controls and how age routing works.
OpenAI's new post says ChatGPT now has 1.2 billion users across many countries. It frames teen access as an equity question: young people with fewer resources should also get help learning and planning, and the product should be designed with that in mind.
The usage numbers, read carefully
OpenAI reports several data points about how teens use the product.
- Learning messages. As ChatGPT for Teens reached more teens globally, those with access sent about 2.7 million more learning-related messages on average than teens who did not yet have it. OpenAI does not state the group sizes in the post, so treat the figure as a direction, not a per-student effect.
- Popular prompts. The most selected conversation starters are "ask questions to develop my ideas," "give feedback on my draft," and "help me check my evidence."
- Tool adoption. In one week, nearly 1.2 million teens used Learning Visualizations for math and science concepts, and more than 180,000 used Study Mode for step-by-step support.
- Time spent. Teens average under 15 minutes a day. Fewer than 2% spend more than three consecutive hours. Among teens with such long sessions, over 80% had at least one learning-related prompt.
- Break reminders. In almost half of the conversations where a reminder appeared, the teen took a break or ended the conversation within five minutes.
OpenAI compares the 15-minute average with a publicly reported average of five hours a day of teen social media use, and argues the product is for "learning, building, and creativity, rather than passive scrolling."
What the numbers do not tell you
This is a company report on its own product, and a few limits are worth keeping in mind.
- An average hides the tail. "Under 15 minutes" is an average. Less than 2% over three hours still means a large absolute number of teens when the base is large. OpenAI does not give the total number of enrolled teens.
- Comparing with social media is not like for like. Social media time is a daily total across many apps. ChatGPT time covers a single product, and teens may use other chatbots too.
- Engagement is not safety. Short sessions say nothing about whether a crisis conversation was handled well, whether parents were alerted, or whether the age system caught the right accounts. Those are exactly the areas the outside test targeted.
- No methodology is published. The post does not describe the sample, the time window beyond "one week" for tool counts, or how teens were identified.
None of this makes the data wrong. It means the report answers "how much do teens use it and for what," not "is it safe."
College Planner: what it does
OpenAI says hundreds of thousands of US teens already use ChatGPT each week to plan for college: tracking deadlines, researching schools and understanding financial aid. College Planner is meant to give that a structure.
According to the announcement, it will:
- Collect application requirements, deadlines, tasks and financial-aid steps for every school on a student's list into a single plan.
- Let students return to update progress and use a timeline to see what is left.
- Let students ask ChatGPT questions in context, from what a requirement means to how to get started.
- Help with financial aid: understanding requirements, tracking each school's aid deadlines, and exploring scholarships and fee waivers.
The first release targets US high-school students in grades 10 to 12 who plan to attend a four-year college. OpenAI says it plans to reach more countries and to cover two-year colleges, technical colleges and trade schools over time. No date is given beyond "soon".
OpenAI also draws a line on scope: the tool is meant to manage the practical steps "without taking over the personal work of figuring out where they want to apply." In practice, how well it keeps to that line, for example on essays, is something only hands-on use will show.
The College Advising Corps tie-in
Alongside the product, OpenAI says it is supporting College Advising Corps, a nonprofit that places recent graduates as advisers in under-resourced high schools. The contribution will help it expand into new states and create an AI fellowship program, in which advisers in participating states will test and share practical ways to use AI. College Advising Corps CEO Ekaterina Struett is quoted in the post saying the tool can help students prepare for conversations with advisers and stay on track.
Flashcards, quizzes and multi-photo capture
The study additions are the part you can reason about most concretely.
- Continuous multi-photo capture. OpenAI says photographing and uploading notes can take hours. On iOS, students can now photograph several pages and have them combined automatically into one PDF ready to upload. Android is in progress.
- Flashcards. A student can upload class notes or choose a topic, and ChatGPT turns the material into cards. Students can mark cards they know, revisit missed ones, or shuffle the deck. Decks are saved in the teen's Library, and a student can ask ChatGPT to schedule a task to practice them.
- Quizzes. Students can upload notes to produce interactive quizzes, and ChatGPT is better at recognizing when someone wants a quiz and creating one directly in chat.
For builders, this is another sign that the chat window is turning into a study surface with state: saved decks, scheduled tasks and interactive widgets. That fits the direction of ChatGPT's Intelligent UI with GPT-6, where answers are interactive objects instead of text.
The teen AI council with Boston Children's Hospital
OpenAI says it is supporting Boston Children's Hospital's Digital Wellness Lab and its independent Student Advisory Council for the next three years. For the 2026 to 2027 school year, about 22 students will take part across two tracks: roughly 12 students aged 14 to 18 working on digital wellbeing projects in their communities, and roughly 10 students aged 16 to 20 focused on building AI chatbots and emerging technology.
Students in the AI track will test tools and recommend changes. OpenAI says it expects to seek feedback on teen safety defaults, parental controls and notifications, and AI literacy, and aims to report back to students on what it changes and why. The post attributes a quote to the Lab's director and executive director that young people "see what works, what fails, and what needs to change."
The council is independent in the sense that the Lab runs it, and the post says OpenAI is funding support. A three-year commitment is useful, but note that the stated aim is to bring recommendations "into product and model decisions early." It is a feedback channel, not a binding oversight body.
How this sits next to the Common Sense Media test
Reading the two items together is the useful exercise.
| Question | OpenAI's October 7 report | Common Sense Media, same day |
|---|---|---|
| Focus | Usage, learning tools, new features | Safeguards: crisis referrals, parent alerts, age prediction |
| Method | Company telemetry, method not published | Roughly 4,000 prompts across two test windows |
| Main message | Short sessions, learning-heavy use | Teen mode safeguards failed in several tests |
| Status | Self-reported | Disputed by OpenAI, not independently replicated |
They do not contradict each other. Teens can spend a short time per day and mostly study, while a small set of high-stakes conversations still goes badly. Neither document alone settles whether the product is safe enough. For the detail on the critical test, see our Common Sense Media coverage. The regulatory backdrop includes the FTC probe of OpenAI, Anthropic and other AI firms.
OpenAI's post closes by saying it will keep "checking how protections work in practice and sharing what we learn." The most useful follow-up would be published methodology and, ideally, access for outside evaluators.
What this means for parents, teachers and builders
Parents. The usage numbers are reassuring on volume but say nothing about content. Link the teen account to your own, check Study Hours and notification settings, and keep talking to your teen. Our guides on AI and parenting and on whether to teach a child AI cover that conversation.
Teachers and counselors. College Planner is aimed at the logistics that overwhelm students without family support. If it works as described, it frees advisers for judgment calls, which is the stated logic of the College Advising Corps fellowship. Expect to check its deadlines against each school's own site, since a planner is only as accurate as its data.
Builders. Three patterns stand out: saved study objects (decks and quizzes in a Library), scheduled tasks tied to learning, and multi-page capture to cut upload friction. If you build ed-tech, the bar for a "chat plus notes" product just moved. Also note how OpenAI discloses safety evidence: engagement stats up front, evaluation methods missing. If you ship to minors, publish both. The OpenAI MentalHealthBench write-up shows what a more open evaluation looks like, and Melo, the explainx.ai learning copilot is our own take on guided learning.
Open questions
- When will College Planner ship, and will it reach teens outside the US soon?
- How many teens are enrolled, and how was "under 15 minutes" calculated?
- Will the teen council's recommendations be published, and which will change defaults?
- Will independent evaluators get access to test the safeguards the Common Sense Media assessment questioned?
We will update this post if OpenAI answers any of these.
Bottom line
OpenAI's report is a useful data point and a product roadmap, not a safety audit. It says teens use ChatGPT briefly and mostly for learning, and it ships real study tools. It leaves unanswered the hard safety questions raised the same day. Read the numbers as claims from the company, and judge the product by whether outside tests start to agree with them.
Details are accurate as of October 8, 2026. College Planner has no launch date yet, and figures come from OpenAI's own post.
Related reading
- Common Sense Media rates ChatGPT for Teens "Unacceptable Risk"
- ChatGPT for Teens: Study Mode, safety controls, and what changed
- ChatGPT Intelligent UI with GPT-6, explained
- FTC confirms a probe of OpenAI, Anthropic and other AI firms
- AI and parenting: a guide for families
- OpenAI MentalHealthBench: scores and criticisms