OpenAI has announced a significant shift in its age verification process for ChatGPT, moving away from the “honor system” to an AI-driven prediction model aimed at identifying users under 18. This update employs behavioral signals—monitoring account activity duration, active hours, and usage patterns—to estimate a user’s age, thereby triggering stricter safety measures for suspected minors. OpenAI acknowledges that while this approach aims to enhance accuracy, it may still lead to misclassifications. Adults misclassified as underage can regain full access through a selfie-based verification with Persona, a third-party service. Privacy advocates are concerned about the reliability of age inference based solely on behavior. Furthermore, the rollout comes amid increasing scrutiny of AI interactions with minors, following reports of inappropriate behavior by AI bots. The new age-prediction model is now active for ChatGPT consumer plans, with an EU rollout planned soon.
Source link
Share
Read more