ChatGPT for Teens is not necessarily 'safer than the previous version,' (www.cnbc.com)

OpenAI's introduction of ChatGPT for Teens, designed to provide a safer AI experience for users aged 13 to 17, faces significant scrutiny from Common Sense Media's Youth AI Safety Institute. Despite being marketed as having "stronger built-in safety protections," recent evaluations indicate that the platform falls short of its safety promises, receiving an "unacceptable risk" level after testing over 4,000 prompts. Experts found that many intended safety features, such as parental alerts for crisis-oriented prompts, were ineffective, raising concerns about young users’ emotional wellbeing when interacting with the chatbot. The report highlights critical discrepancies between OpenAI's claims and the operational effectiveness of the safety features designed for teens, such as the refusal of explicit content and the bot’s implications of having emotions or preferences, which can foster unhealthy attachments. While OpenAI acknowledges the findings and welcomes independent evaluations, they contend that some results may stem from earlier phases of feature activation. The tension between ensuring a safe user experience and maintaining a viable business model remains a contentious issue as the AI landscape continues to evolve.