OpenAI Debuts ChatGPT for Teens as Safety Experts Advise Caution

OpenAI has launched ChatGPT for Teens, a specialized version of its flagship generative artificial intelligence chatbot designed for younger users. Positioned by the San Francisco-based startup as an educational partner, the product aims to limit emotional and intimate interactions between adolescents and software. The rollout follows intense public scrutiny and lawsuits alleging that heavy engagement with ChatGPT contributed to youth self-harm and suicide.

Child safety advocates and educational experts urged caution following the release, telling Mashable that independent evaluation is required before trusting the tool's safety mechanisms. Robbie Torney, head of AI and digital assessments at Common Sense Media’s Youth AI Safety Institute, told Mashable that while OpenAI has made notable safety commitments, third-party testing must confirm their real-world performance before caregivers place full confidence in ChatGPT for Teens.

Torney emphasized that teenagers will not necessarily be placed into the restricted experience automatically. OpenAI utilizes age estimation algorithms to sort users into age-appropriate accounts, but the system can misclassify minors as adults—especially if a minor previously provided a false birth date. To ensure a child accesses the teen-specific edition, parents must link their account to their child's profile and explicitly enable parental controls.

The platform's management controls allow caregivers to block sensitive material, set Quiet Hours to disable access entirely, and activate Study Mode, which restricts outputs to learning content. Enabling these settings is also necessary for parents to receive alerts if a teenager discusses self-harm, suicide, or violence. However, Torney noted that previous evaluations by the Youth AI Safety Institute revealed notification alerts were delayed by one to two days or failed to arrive altogether. An OpenAI spokesperson stated that trained human reviewers examine every safety alert before dispatch, aiming for a turnaround time under one hour.

Lauren Jonas, head of youth and families at OpenAI, told Mashable that ChatGPT for Teens is built to redirect personal or emotional conversations back toward task-oriented study. However, preliminary testing by the Youth AI Safety Institute showed that researchers were still able to engage the system in personal support conversations unrelated to academics. Camille Carlton, senior director of strategy and impact at the Center for Humane Technology, questioned the fundamental premise of AI companionship for youth, stating that critical developmental capacities are formed through real-world offline experiences rather than interactions with machines.

Questions also remain regarding the educational efficacy of the software. Dr. Hamsa Bastani, associate professor of Operations, Information, and Decisions at the Wharton School of the University of Pennsylvania, previously characterized OpenAI's initial effort at homework help as weak, noting that many commercial AI tutors were simply standard large language models wrapped in basic anti-cheating prompts. While Bastani's research group published its first positive AI learning findings five months ago, she has not yet tested the upgraded teen platform. Jonas stated that OpenAI is currently executing randomized controlled trials for Study Mode across multiple countries to measure learning outcomes.

Industry observers warned that frictionless AI assistance can inadvertently discourage genuine skill building. Dylan Arena, chief data science and AI officer at McGraw Hill, noted that general-purpose chatbots can lead teenagers to over-rely on automated solutions instead of developing their own problem-solving skills and navigating social dynamics. A recent Common Sense Media report similarly found that a majority of students lean on AI for direct answers even when attempting to use chatbots appropriately. Furthermore, research co-authored by Bastani found that poor self-regulation causes students to over-rely on AI assistance even when aware it harms learning, raising doubts about whether teens can self-regulate effectively when frustrated by schoolwork.