OpenAI is arguing that teenagers should have access to AI rather than being kept away until adulthood, provided the products include protections designed for younger users. The company says nearly nine in ten teens who used ChatGPT in a single week used it for learning, information, skill-building, or productivity. That figure is OpenAI’s own measurement and should be read as such.

The post combines an access argument with a product-safety update. It points to stronger default protections for accounts estimated to belong to someone under 18, age prediction, linked parental controls, family resources, and learning-oriented experiences. OpenAI frames the goal as helping teenagers learn and create while limiting exposure to material inappropriate for their age.

The controls OpenAI describes

  • Parents with linked teen accounts can enable Study Mode so new chats begin in a learning-focused experience.
  • Teen experiences receive stronger safeguards around self-harm, graphic violence, risky challenges, unhealthy body-image content, and dangerous or sexual roleplay.
  • Parental tools can manage selected features and usage patterns, while family resources explain safer use.
  • OpenAI says interactive math and science experiences have expanded to more than 300 topics, with 18 million weekly users across all ages engaging with them.

Policy claims are not outcome evidence

A list of safeguards does not tell parents how often they work, how often they block harmless requests, or what happens when a teenager is incorrectly classified as an adult. Age prediction is especially consequential because it determines which defaults apply. Useful transparency would include error rates across regions and devices, appeal pathways, and independent testing against realistic attempts to bypass protections.

Learning features require similar care. A mode that asks guiding questions can support active practice, but a conversational system can also encourage overreliance or confidently present mistakes. Schools and families need controls that are understandable, not merely available, plus clear limits on when an AI assistant should give way to a teacher, parent, counselor, or emergency resource.

The regulatory issue is moving from whether teenagers use generative AI to how companies identify younger users and prove age-appropriate design. That includes default settings, content policies, data handling, advertising boundaries, crisis escalation, and parental oversight. Different countries may set different thresholds, making the evidence behind these systems as important as the feature list.

OpenAI’s post is valuable as a statement of its commitments and current product direction. It should not be mistaken for an independent conclusion that teen safety has been achieved. The next meaningful reporting should test the controls and compare the company’s stated design with measurable outcomes.