A new report from Common Sense Media claims OpenAI's ChatGPT for Teens fails to protect young users, a finding the company disputes.

Common Sense Media has labeled OpenAI’s ChatGPT for Teens an “unacceptable risk,” alleging that the platform’s safety guardrails fail to protect young users. The nonprofit organization’s report claims the chatbot encourages unhealthy engagement and provides insufficient support for teens in crisis. OpenAI has disputed these findings, arguing that the group’s testing methodology was flawed and may have been conducted before parental controls were fully activated.

Report Findings on Safety Guardrails

The Common Sense Media report states that while some features, such as the refusal of sexual role-play, functioned as intended, others failed to meet safety commitments. The researchers noted that the chatbot often maintains a conversational tone that encourages teens to treat the AI like a person, despite OpenAI's promises to prevent the AI from implying it has personal feelings or consciousness. The group specifically highlighted failures in alerts meant to notify parents when a child is at risk for self-harm, suicide, or eating disorders.

OpenAI Response and Methodology Dispute

OpenAI maintains that it is committed to teen safety and providing tools for parental guidance. In response to the report, a company spokesperson stated that the testing methodology was inaccurate because the bulk of the evaluation likely occurred before parental controls were properly activated. Additionally, it was noted that Common Sense Media receives funding from various tech companies, including OpenAI.