ChatGPT for Teens Labeled 'Unacceptable Risk' by Common Sense Media

Common Sense Media has released a detailed assessment declaring OpenAI’s ChatGPT for Teens an “unacceptable risk” due to systemic failures in safety design and parental communication.

The organization’s review found that the teen-focused interface fails to trigger essential alerts when children encounter harmful content, notably in cases involving self-harm where parents receive no notification despite extended conversations. Additionally, the system continues to assist with academic tasks, effectively doing kids’ homework, which undermines educational objectives.

Executive Director Tom Siegel warned that these shortcomings create false confidence for parents and could enable dangerous behavior to go unnoticed for hours. He urged OpenAI to fix these issues through independent verification before releasing updates.

OpenAI** has countered by defending its approach, stating that its testing methodology was compromised—specifically, that parental control activation took several hours in some cases, rendering early assessments inaccurate. The company emphasized ongoing collaboration with experts to enhance teen learning features while maintaining robust safety mechanisms.

From an infrastructure perspective, this debate underscores the broader challenge of aligning advanced AI models with responsible deployment practices across cloud GPU platforms. As AI-powered applications become integral to education and daily life, ensuring reliable safety rails becomes a critical requirement for any AI-as-a-service offering deployed on scalable compute infrastructure.