ChatGPT for Teens presents 'unacceptable risk' (foxbaltimore.com)

OpenAI's newly released ChatGPT for Teens is facing criticism for posing what has been deemed an "unacceptable risk" to underage users, according to a recent report from the Common Sense Media Youth AI Safety Institute. Conducting tests on over 4,000 prompts pre- and post-launch, the institute found that the chatbot's safety features fell short of expectations, notably in its parental notification system, which failed to alert guardians about concerning discussions on mental health or self-harm. This suggests that the alerts depend more on account history than the severity of the content, raising serious concerns about the effectiveness of the protective measures intended for young users. Further compounding these issues, the report highlighted that ChatGPT for Teens could be easily manipulated by students seeking to bypass academic integrity systems, potentially undermining the educational experience. While OpenAI maintains that it is committed to teen safety and argues that the testing methodologies might not accurately reflect the platform's safety measures, the findings have sparked a call for a pause in marketing the product until robust, developmentally appropriate safety solutions are implemented. This incident underscores the complexities of developing AI tools for youth, where the balance between advanced technology and user safety is critically essential.