San Francisco:Some safety measures implemented in OpenAI's ChatGPT for Teens are reportedly ineffective, posing what is described as an 'unacceptable risk' to children, as detailed in a report by Common Sense Media released on Wednesday.
According to United News of Bangladesh, the watchdog group has urged OpenAI to restrict the chatbot's access to adults only until a secure and developmentally suitable experience can be assured. In response, OpenAI contended that the testing conducted by Common Sense Media does not accurately represent the effectiveness of their safety measures.
Concerns have been expressed by parents, educators, and child development experts regarding children's use of AI chatbots. Issues raised include the potential for students to use these tools to cheat on schoolwork and, in extreme cases, the contribution of such tools to incidents of suicide. Experts also warn that teenagers, whose brains are still in development, may be particularly susceptible to forming unhealthy attachments to AI chatbots.
In its report, Common Sense Media indicated concerns that ChatGPT for Teens might give parents misplaced confidence in the efficacy of safety features that often fail. While some teen-mode features, like the chatbot's refusal to engage in explicit sexual role-play, functioned correctly, other safeguards did not. Alerts intended to notify parents of risks such as suicide, self-harm, and eating disorders reportedly failed, as did measures aimed at supporting teens in crisis.
The report also highlighted that ChatGPT can still converse in a manner resembling a friend when teens interact with it as if it were a person. OpenAI has previously stated that the chatbot would not imply it possesses personal feelings, consciousness, or emotions.
OpenAI maintains that it is 'deeply committed to teen safety' and is actively working to enhance its safeguards while equipping parents with tools to oversee their teenagers' use of AI. The company also noted that a review of the report's methodology suggested that much of the testing may have occurred before the full implementation of parental controls, thus questioning the report's accuracy.
Common Sense Media, which advocates for responsible digital media use, receives funding from OpenAI and other technology companies, including some whose products it evaluates.