OpenAI strengthens ChatGPT safeguards for users aged 13 to 17: Here's how

OpenAI has unveiled significant updates to its internal guidelines governing the behavior of its artificial intelligence systems, specifically aimed at protecting users aged 13 to 17. The San Francisco-based organization emphasized that the safety of teenage users now takes precedence over prior objectives, including maximizing helpfulness and user freedom. This decision comes in response to increasing concerns regarding how generative AI technologies impact younger audiences. In a recent blog post, OpenAI acknowledged that teenagers have distinct developmental needs that necessitate a more cautious and age-appropriate approach from AI systems. While the foundational principles of the Model Spec remain applicable to all users, the updated guidelines clarify how these principles should be applied in interactions involving minors, particularly during sensitive or potentially high-risk discussions. Under these revised standards, ChatGPT is instructed to prioritize the safety of teenagers, even if it means conflicting with other established goals. This marks a considerable shift in the platform's approach, as the protection of minors now supersedes its previous focus on openness and flexibility. OpenAI has outlined four key principles that will shape ChatGPT's interactions with teenage users. The chatbot is now tasked with promoting real-world support, encouraging connections with trusted adults and offline relationships rather than serving as a substitute. It must engage with teenagers appropriately, avoiding condescension and the assumption that they can be treated as adults. Furthermore, transparency is a focus, with ChatGPT expected to set clear expectations regarding its capabilities in conversations tailored for teens. The system will also be less likely to offer flattery or excessive agreement, as OpenAI has adjusted its sycophancy metric for this age group. Additionally, the company announced that it is in the early stages of rolling out an age prediction model, which will analyze subtle conversational cues to assess whether a user is likely under 18, even if they haven’t shared their age. This new feature is expected to be gradually implemented across various ChatGPT consumer plans soon.