OpenAI says teen ChatGPT use limited but research finds it an ‘unacceptable risk’

OpenAI says teenagers are generally using ChatGPT for less than 15 minutes a day, but a new safety assessment has raised serious concerns about whether the platform is safe enough for younger users. Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “unacceptable risk” after testing more than 4,000 prompts and identifying gaps in parental alerts, crisis responses, and learning protections. OpenAI has disputed the findings and said the testing may not accurately reflect how its latest safeguards work. The disagreement highlights growing concerns over AI safety, teen mental health, parental oversight and responsible technology use.

ChatGPT for Teens and Safety Concerns

OpenAI has introduced ChatGPT for Teens with a focus on helping teenagers use artificial intelligence and an AI chatbot in a healthier way. The service includes safety features, stronger safety guardrails and tools intended to support users aged 13 and 17, with attention to emotional dependence, problematic use and other concerns involving young users.

The teen-focused features include parental alerts, parental controls, learning activity and support for study help and school work. OpenAI has also said that average teen use is less than 15 minutes a day, while automatic reminders can encourage users to take a break after longer stretches of use, including three consecutive hours of activity.

Learning prompts and other safeguards are intended to encourage healthy use rather than excessive engagement. An OpenAI spokesman has described the measures as part of a wider effort to make ChatGPT safer for younger users while maintaining access to educational and everyday assistance.

Teen Guardrails and Parental Controls

ChatGPT for Teens includes teen users within a framework of parental controls, allowing parents to create or manage linked accounts and receive certain notifications. The safeguards address sensitive areas including self-harm, disordered eating, romantic language and emotional dependence, with additional measures intended to protect young people.

The system is designed to be conscious of situations involving sexualized imagery and sensitive image content. Safety features, reminders and healthy use tools form part of the additional controls, while teen safeguards are intended to reduce risks associated with prolonged or inappropriate chatbot interactions.

Safety alerts, problematic use monitoring and parent-linked accounts can also support parental notification in situations covered by the safeguards. These safety measures are part of OpenAI’s broader approach to artificial intelligence and child protection, although their effectiveness remains an important subject of discussion.

Common Sense Media Research Findings

Common Sense Media has published new research warning of an unacceptable risk to young people and questioning whether current teen guardrails provide reliable safeguards for people under 18 years old. The research examined platform access, parental controls and the way safeguards performed across different testing conditions.

Its methodology included testing multiple accounts, including teen accounts, newly created accounts and parent-linked accounts, as well as accounts with accumulated conversation history. Researchers examined interactions involving sensitive topics, sexual roleplay, young users, suicide, suicidal ideation, self-harm and disordered eating.

The evaluation also examined parental alerts and found zero alerts in certain tests involving older accounts with weeks of conversation history. The findings were presented as an independent evaluation of AI safety, raising questions about whether safeguards work consistently across different account histories and sensitive situations.

Self-Harm Alerts and Crisis Support

Self-harm, suicide, suicidal ideation and disordered eating are among the areas examined in relation to ChatGPT for Teens and its parental alerts. In testing involving newly created accounts and parent-linked accounts, researchers examined whether safety alerts were triggered after as much as one hour of conversation, including interactions involving sensitive topics.

The reported zero alerts in some tests raised questions about the reliability of parental notification when a crisis develops within a conversation. The research also considered the role of a hotline, professional support and crisis-related recommendations when young people discuss serious mental-health concerns.

Reliable safeguards remain central to teen safety, particularly because safety failures can create false confidence among parents. ChatGPT conversations, mental health concerns and the experiences of young people have therefore become important areas of examination for groups such as Common Sense Media, Tom Siegel and the Youth AI Safety Institute.

ChatGPT Learning Use and Anthropomorphic Behaviour

OpenAI says teen ChatGPT use often involves learning, study help and school work, with learning prompts supporting educational activity. At the same time, OpenAI has highlighted patterns in which users may engage for three consecutive hours, while saying that more than 80% of the average teen user spends less than 15 minutes a day using ChatGPT.

Break reminders and automatic reminders can appear during longer sessions, encouraging users to stop using ChatGPT. These measures are intended to support learning activity, educational use and schoolwork without encouraging excessive engagement, particularly among young users.

Concerns also extend to sexual roleplay, anthropomorphised language and situations where users talk like it’s a friend. Questions about emotional dependence, whether an AI system appears conscious, and the effectiveness of safety guardrails remain part of the broader debate over teen safeguards and chatbot behaviour.

AI Platforms and Legal Pressure

Young people are increasingly at the centre of debates about online platforms, self-harm, suicide, children and mental health. Companies including Meta and TikTok have faced lawsuits and growing scrutiny over alleged harms involving young users, while technology companies have faced pressure over mental health risks and online safety.

The concerns surrounding OpenAI and ChatGPT have added another dimension to this debate. Reports connected to the Tumbler Ridge mass shooting in rural Canada in February involved an 18-year old and questions about ChatGPT conversations, law enforcement and the role of AI in the months before the incident. OpenAI issued an apology in connection with the reported examination of the shooter’s account, while several lawsuits have increased attention on AI safety.

The wider debate now involves AI safety, child safety, social media platforms, mental health risks, technology companies and the responsibility of companies developing AI systems. As scrutiny and legal pressure increase, the effectiveness of safeguards for young users remains a central issue for parents, researchers and companies.

Conclusion

The debate over ChatGPT for teens highlights the difficult balance between giving young people access to useful AI tools and protecting them from potential risks. OpenAI points to relatively short daily usage and features such as parental controls, break reminders, and learning-focused tools as evidence of its efforts to encourage safer use. However, Common Sense Media’s assessment argues that important safeguards did not consistently work during its testing, particularly around crisis referrals and parental notifications. As AI becomes increasingly common in education and everyday life, independent testing, stronger protections, and clear parental oversight will remain important parts of the conversation about teen AI use.

FAQs