OpenAI Says Teen ChatGPT Use Is Limited, but Research Flags 'Unacceptable Risk'

OpenAI has said that use of ChatGPT by teenagers is limited, even as new research cited by the BBC concluded that the chatbot poses an "unacceptable risk" to younger users due to failures in its safety guardrails.

The research reportedly found that protections designed to shield teens from inappropriate or harmful content and interactions frequently did not function as intended, raising concerns among child safety advocates and researchers about the adequacy of current safeguards on widely used AI chatbot platforms.

OpenAI's own characterization that teen usage is limited appears to be offered as context around the scale of potential exposure, though the company has not disputed the substance of the findings regarding guardrail failures in the available reporting. Specific details on the methodology of the research, including sample size or how guardrail failures were tested, were not fully outlined in the report.

The findings add to a broader, ongoing debate over the safety of AI chatbots for younger users, as companies including OpenAI face increasing scrutiny over how their products interact with minors, particularly around sensitive topics such as mental health, self-harm and age-appropriate content. Several AI companies have introduced or strengthened age-verification and content-moderation systems in response to similar concerns in recent years.

OpenAI has previously said it is working to improve safety systems for younger users and to refine how its models detect and respond to users who may be minors. The company has not detailed specific new measures in direct response to this particular research as of the report.