• OpenAI claims teens average under 15 minutes daily on ChatGPT, with fewer than 2% using it for more than three consecutive hours.
  • The company says users of ChatGPT for Teens sent 2.7 million more learning-related messages than non-users and is rolling out a College Planner.
  • Common Sense Media's Youth AI Safety Institute on October 7 labeled the platform an 'unacceptable risk,' citing missed crisis referrals; OpenAI disputes the testing methods.

OpenAI is pushing back against a scathing safety assessment with new usage statistics, claiming that teenagers spend less than 15 minutes a day on ChatGPT on average and that fewer than 2% use the platform for more than three consecutive hours. The company also says that users of ChatGPT for Teens sent an average of 2.7 million more learning-related messages than non-users, and it is preparing to launch a College Planner as students begin the application process.

The numbers, which could not be independently corroborated as of October 7, 2026, 12:01 p.m. EDT, come as OpenAI faces mounting scrutiny over how its artificial intelligence tools affect young people. The usage metrics describe different measures—average daily time versus prolonged sessions—and leave unanswered critical questions about the measurement period, the population covered, and the distribution of use.

The learning claim also requires clarification. The phrase '2.7 million more learning-related messages on average than non-users' has an unclear denominator and comparison group. It should not be interpreted as an average increase per teenager or as evidence of improved academic performance without the underlying methodology.

A Competing Safety Assessment

The most important development, however, is not OpenAI's usage data but a competing safety evaluation. On October 7, Common Sense Media's Youth AI Safety Institute tested more than 4,000 prompts on accounts registered to ages 13–17 and called ChatGPT for Teens an 'unacceptable risk.' The institute found that while ChatGPT generally avoided instructions facilitating self-harm and other prohibited content, it missed more than one in four situations its testers judged to warrant a crisis referral. More than a dozen newly created, parent-linked accounts also produced no parental alert during conversations lasting up to an hour about suicide, self-harm, or disordered eating.

'These announced protections have industry-leading potential but fall short in practice,' said Tom Siegel, head of the Common Sense institute, warning against outsourcing care for children to an immature product.

OpenAI disputed the testing methods, saying parental-account linking can take several hours, potentially invalidating some alert tests. The company also says age prediction can take up to two weeks, with intermediate protections during that period, and that allowing teenagers to exit Study Mode during Study Hours is intentional flexibility rather than a defect.

The central issue remains safety: low average usage and more learning-related messages do not, by themselves, establish that safeguards work for vulnerable teenagers. A low population average can coexist with serious problems in a small subgroup, and time spent is not a sufficient safety measure.

OpenAI did not immediately respond to a request for comment on the Common Sense findings. A spokesperson previously said the company's own crisis-resource data shows significantly higher referral rates than the institute's tests indicate.

The Financial and Regulatory Landscape

The dispute unfolds against a backdrop of rapid commercial expansion. OpenAI reported more than 900 million weekly active ChatGPT users and over 50 million subscribers as of March 31, though those are dated scale indicators, not verified October totals. Reuters (TRI) reported on September 29 that annualized recurring revenue was approaching $70 billion, with enterprise sales more than doubling since July—a revenue run rate, not revenue actually earned during 2026 or evidence of profitability.

OpenAI announced $122 billion in committed capital on March 31 at an $852 billion post-money valuation. But Reuters also summarized an FT report projecting $278 billion in cash burn during 2026–2030, underscoring that rapid revenue growth still accompanies enormous expected funding needs.

Regulatory pressure is intensifying. Fortune reported on September 14 that California enacted 'Adam's Law,' requiring age verification, parental controls, crisis support, restrictions on targeted advertising to children, and measures against harmful or emotionally manipulative outputs. The law followed the 2025 death of Adam Raine and allegations about his ChatGPT interactions, which should not be treated as an adjudicated finding of responsibility.

OpenAI has shifted from warning against fragmented state rules toward helping shape state legislation while continuing to favor a national framework. Policy analyst James Czerniawski argues that federal inaction encourages state intervention but warns that geographically fragmented rules and VPN circumvention complicate enforcement.

Internationally, no direct diplomatic or cross-border dispute is established by this headline. The broader issue is whether multinational AI platforms adopt common youth protections or different regional experiences—a prospective compliance question, not a demonstrated international consequence here.

Education Push and Staffing Turmoil

OpenAI is expanding beyond general chat into structured learning. Its August teen launch combined Study Mode, quizzes, learning visualizations, homework reminders, and Study Hours. A College Planner would fit that broader direction, although its launch date, availability, pricing, and capabilities could not be verified and should be described as forthcoming rather than already available.

Potentially, this could lower the cost of supplementary tutoring and application guidance while increasing competitive pressure on education-service providers. Those are plausible consequences, not measured effects of this announcement.

Meanwhile, the company faces internal turbulence. Fortune reported on October 5 that safety-report lead David Robinson resigned and three researchers were dismissed for alleged confidential-information violations. The underlying circumstances remain incompletely disclosed, intensifying debate over independent scrutiny and safety transparency. Sam Altman remains CEO; the retrieved reporting does not identify a CEO change tied to this teen-use announcement.

A useful parallel is Anthropic's different access policy: Fortune reports that it does not allow under-18 users and therefore negotiated exclusion from the California bill. Google (GOOGL), Meta (META), and Amazon (AMZN) also participated in legislative discussions, showing that youth-chatbot regulation extends beyond OpenAI.

What to Watch

Short term, expect pressure for clearer activation notices, reproducible independent tests, and more transparent reporting of crisis referrals and parental alerts. Resolving the methodological disagreement matters more than publishing another average-use statistic.

Long term, teen-focused AI could broaden access to educational support, but adoption will depend on demonstrating both learning outcomes and reliable protection for higher-risk users. Expert perspectives remain opposed rather than convergent: Siegel emphasizes failures and the need for real-world support, while OpenAI's launch materials quote educator Racquel Gibson on the potential of guarded, step-by-step assistance for students without homework help at home.

Neither perspective independently verifies the headline's learning statistic. Related developments include OpenAI's August partnership with CodeAI for student and educator AI literacy, its campus-focused Student Collective launched in April, and the recent safety-team departures. Together, these show simultaneous expansion into education and rising pressure to substantiate safety claims through external evaluation.

Correction: An earlier version of this article misstated the date of Common Sense Media's assessment. It was October 7, 2026, not October 7, 2025. Additionally, the headline's precise usage statistics and College Planner announcement could not be independently corroborated in the sources retrieved; they are treated as company claims.