OpenAI’s ChatGPT for Teens,
launched in August 2026 to curb AI-related risks for users aged 13 to 17, is facing scrutiny over its effectiveness. Recent testing by Common Sense Media reveals failures in crisis resource referrals and parental notification systems, raising concerns about the tool’s safety features for vulnerable students.
Safety Gaps in Crisis Management
While OpenAI designed its teen-specific mode to limit high-risk conversations involving self-harm, violence, eating disorders, and graphic content, independent analysis suggests these safeguards are inconsistent. Common Sense Media, a nonprofit that evaluates technology for families, conducted an exhaustive study using 4,000 prompts to test the chatbot’s responses before and after its official launch.
The findings indicate that the chatbot frequently fails to direct teenagers toward professional help. In instances involving potential psychosis, disordered eating, or suicidal ideation, the system did not consistently recommend crisis hotlines. According to Common Sense Media, the chatbot failed to provide referrals to crisis resources 25% of the time in high-harm scenarios where such guidance was warranted.
Parental Control Failures
The product’s architecture includes parental controls designed to alert guardians when a teenager discusses sensitive topics. However, testing revealed significant technical lapses in these features. When researchers used newly created, parent-linked accounts to engage in hour-long conversations regarding self-harm or suicidal ideation, OpenAI sent zero notifications to the linked parent accounts.
Even on older, established accounts, notifications were not triggered by sensitive talk until after weeks of interaction. Consequently, Common Sense Media has called for the feature to be suspended, arguing that marketing safety tools that do not function as promised is misleading. You shouldn’t be marketing safety features to parents and young people that don’t work. We believe it should be paused,
the nonprofit stated.
Chatbot Generates Essays Despite Study Mode Constraints
Beyond safety concerns, the tool’s impact on education remains a primary worry. Despite the inclusion of a “study mode” intended to guide students through homework rather than providing direct answers, testers found it trivial to bypass these constraints. When prompted to write essays on literary works like Heart of Darkness or Dubliners, the chatbot consistently generated complete, eloquent compositions.
This behavior highlights the challenge of preventing students from using the tool as a shortcut for complex schoolwork, which studies suggest may stifle the development of critical thinking and information retention.
For those in need of support, resources include the Mental Health Psychosocial Support Service at 03-2935 9935 or 014-322 3392, Talian Kasih at 15999 or 019-261 5999, and Befrienders Kuala Lumpur at 03-7627 2929.
