 (3).png)
1 month ago
3
ARTICLE AD BOX
OpenAI has updated its Model Spec to prioritise the safety of teenagers using ChatGPT. The new rules place protection for users aged 13 to 17 above other goals, introduce age-appropriate behaviour, and expand age detection tools to create a safer AI experience.
OpenAI on Thursday announced changes to its Model Spec, the internal rulebook that guides how its artificial intelligence systems behave, with a new and explicit focus on protecting teenagers who use ChatGPT.
The San Francisco-based company said the updated guidelines put the safety of users aged 13 to 17 above all other objectives, including the long-standing goal of maximising helpfulness and user freedom. The move comes amid growing scrutiny of how generative AI tools affect younger users.
Teen safety placed above all other goals
In a blog post released on Thursday, OpenAI said teenagers have different developmental needs compared to adults and that AI systems must respond to them in a more cautious and age-appropriate manner.
While the broader principles of the Model Spec continue to apply to everyone, the company said the update clarifies how those principles should be interpreted when minors are involved, especially in sensitive or high-risk conversations.
Under the revised rules, ChatGPT will prioritise teen safety even if it conflicts with other objectives set out in the document. This represents a notable shift, as safety for minors now overrides the platform’s usual emphasis on openness and flexibility.
New rules for interacting with younger users
OpenAI outlined four specific principles that will guide how ChatGPT interacts with teenagers.
The chatbot will be required to encourage real-world support, including trusted adults and offline relationships, rather than positioning itself as a substitute. It must also treat teenagers as teenagers, avoiding both condescension and the assumption that they can be handled like adults.
The company also stressed transparency, saying ChatGPT should clearly set expectations about what it can and cannot do in teen-focused conversations. In addition, the system will be less prone to flattery or excessive agreement, as OpenAI confirmed that its so-called sycophancy metric will be lowered for users in this age group.
Age detection tools still being rolled out
OpenAI also revealed that its age prediction model is in the early stages of deployment. This system analyses subtle cues in conversations to estimate whether a user may be under 18, even if they have not disclosed their age.
According to the company, the model will gradually be rolled out to ChatGPT users across different consumer plans in the near future.
Key Takeaways
- OpenAI's new guidelines prioritize safety for teenagers using AI.
- The chatbot will promote real-world support instead of substituting human relationships.
- Transparency in AI interactions with teens is emphasized, including clear expectations.
English (US) ·