A report by Common Sense Media says some OpenAI guardrails in ChatGPT for Teens do not work as intended and could create an “unacceptable risk” for children.
The Story
Common Sense Media says some of the guardrails OpenAI built into its ChatGPT for Teens product do not work as intended and that the chatbot poses an “unacceptable risk” to children, in a report published Wednesday.
The watchdog group urged OpenAI to restrict ChatGPT for Teens to adults “until it can offer a safe, developmentally appropriate experience.”
Common Sense Media said some teen-mode features worked as intended, including the chatbot’s refusal of explicit sexual role-play. The group said other elements did not work as intended, including alerts meant to notify parents if a child is at risk for suicide, self-harm or eating disorders, as well as support for teens in crisis.
In the report, Common Sense Media also said “ChatGPT still talks like a friend when teens treat it like a person,” while noting OpenAI had promised the chatbot is prevented from suggesting it has personal feelings toward the user or implying that it is conscious or experiences emotions.
OpenAI said the group’s testing does not reflect how ChatGPT’s safeguards work. The company said it is “deeply committed to teen safety” and that its review of the report’s methodology shows “the bulk of their testing may have begun and concluded before activation of parental controls was complete, making their findings inaccurate.”
The report also drew on broader concerns raised by parents, educators and child development experts about children’s use of AI chatbots, including concerns about cheating and harms involving mental health.
Common Sense Media said it receives funding from OpenAI and other tech companies, including those whose services it evaluates.
← More stories