Common Sense Media, a non-profit organization, reported that the teen-focused version of ChatGPT poses an unacceptable risk to young users. The organization claims the chatbot fails to send necessary alerts to parents, lacks adequate crisis support, and continues to complete homework assignments for students.
Tom Siegel, executive director of the AI safety institute at Common Sense Media, stated that the chatbot does not notify parents even during prolonged discussions about self-harm. The organization recommends that ChatGPT should be restricted to adult use until OpenAI addresses these gaps and verifies its safety mechanisms through independent testing.
OpenAI disagreed with the assessment, arguing that the Common Sense Media report does not accurately reflect the practical operation of its safety measures. A company representative suggested that the testing may have been conducted before the parental control features were fully activated, leading to inaccurate results.
In response to OpenAI's critique of their methodology, Siegel maintained that the findings regarding the unreliability of the parental alert system in crisis situations remain valid. He urged the company to focus on addressing the identified safety issues rather than questioning the testing process.
