What happened
OpenAI introduced updated AI model guidelines for users under 18, prioritising teen safety by adjusting chatbot core functions. A new 'Model Spec' provides specific guidance for ChatGPT, steering younger users from sensitive topics and encouraging real-world support. OpenAI also released AI literacy resources for families and is testing an age estimation system to automatically activate teen safety features, including parental controls and restrictions on sensitive topics.
Why it matters
The introduction of an age estimation system and dynamic model behaviour based on user age introduces a new dependency on the accuracy and integrity of age verification mechanisms. This creates a visibility gap for platform operators regarding the precise application of content restrictions, increasing exposure for compliance teams to potential policy mismatches if age estimation fails or model steering is circumvented. IT security faces increased due diligence requirements for validating the robustness of the age estimation system.
