OpenAI is rolling out new safety guardrails for ChatGPT designed specifically around how the AI should interact with teenage users, the company announced. The update adds Under-18 Principles to the platform's Model Spec that spell out rules for age-appropriate responses grounded in developmental science.
The new framework clarifies how ChatGPT should behave when engaging with minors, particularly in high-risk scenarios. OpenAI says the changes strengthen existing safeguards and represent an expansion of its broader teen safety efforts across the platform.
The guardrails codify expectations for how the AI assistant should support young users while maintaining appropriate boundaries. The principles draw on developmental research to tailor guidance that accounts for how teenagers process information and make decisions differently than adults.
This move reflects growing industry pressure to protect minors from AI-related harms, from exposure to harmful content to interactions that could encourage risky behavior. By building safety standards into the model itself rather than treating them as afterthoughts, OpenAI is attempting to prevent problems at the source.
The company has faced scrutiny over how its tools are used by underage users, making this update a concrete step toward addressing those concerns. How effectively these principles work in practice will likely become a benchmark for evaluating OpenAI's commitment to teen safety going forward.
Author Emily Chen: "Embedding safety rules into the model spec is the right approach, but the real test is whether these principles actually hold up when millions of teenagers start pushing against them."
Comments