How does "ChatGPT" estimate your age
The article reviews the age estimation mechanism adopted by OpenAI for ChatGPT users, aimed at protecting teenagers. The mechanism relies on multiple indicators such as conversation topics and usage times, without disclosing their relative weights or classification accuracy. It also explains procedures for correcting misclassification through the "Persona" identity verification service.
The article discusses the age estimation mechanism developed by OpenAI for users of the ChatGPT chatbot, aimed at enhancing the protection of teenagers from inappropriate content. The mechanism does not rely on a single indicator but combines five main indicators: the nature of topics discussed by the user, typical usage times, account usage patterns and activity frequency, the age of the account itself, and the age declared by the user. The company emphasizes that none of these indicators alone means the user is a minor, and it does not disclose the detailed weights of each indicator or the overall estimation accuracy. If a user is classified as a minor, they are moved to the teenage experience, which imposes restrictions on violent or sensitive content and provides parental control tools. Adult users who are misclassified can correct this through account settings, using the "Persona" age verification service, which may require a selfie or a government-issued ID. The mechanism raises questions about classification accuracy and privacy protection, especially given the lack of disclosure about the system's accuracy.