In recent discussions surrounding artificial intelligence, the safety of adolescent users has emerged as a paramount concern. As part of its commitment to addressing this issue, OpenAI has collaborated with organizations such as Common Sense Media and everyone.ai to develop comprehensive safety policies tailored for young users.
OpenAI is making these policies available as open source, intending to promote wider implementation and evolution within the AI community. Developers can modify the prompts for their specific applications, expand upon them to address additional risks, and adapt them to various contexts.
Input from external partners underscored the necessity for more systematic approaches to safety.
Robbie Torney, Head of AI & Digital Assessments at Common Sense Media, remarked, “A significant gap in teen AI safety has been the absence of clear, actionable policies for developers to utilize. Developers often start from scratch, but these prompt-based policies establish a vital safety baseline across the ecosystem. Their open-source nature allows for continuous adaptation and improvement. We’re optimistic that this infrastructure will promote shared standards for youth safety throughout the industry.”
Dr. Mathilde Cerioli, Chief Scientist at everyone.ai, added, “Efforts like this that enhance youth safety policies are crucial, as they bridge the gap between expert insights and practical guidance for real-world systems. Content policies represent an essential initial effort and pave the way for further exploration of how model behaviors can influence youth-related risks over time. Motivated by this initiative and our research, everyone.ai has developed a foundational behavioral policy addressing issues such as exclusivity and overreliance.”
Part of a Broader Shift Toward Built-in Safeguards
This release is part of a larger suite of safety initiatives from OpenAI in the past year, which includes revisions to its Model Spec to incorporate protections for users under 18, as well as features like parental controls, age prediction systems, and regional Teen Safety Blueprints.
Lehane emphasized the importance of integrating safety measures early in the development process, stating, “Robust safeguards for teens should be embedded from the outset rather than added as an afterthought.”
He also noted, “We’ve implemented safeguards such as parental controls and age prediction, released a series of Teen Safety Blueprints in various countries, and are collaborating with lawmakers, parents, and educators to establish clearer standards and more comprehensive protections for teens interacting with AI.”
These tools are designed to work within a layered safety framework, urging developers to combine prompt-based policies with thoughtful product design, effective monitoring systems, and user controls.
Lehane reminded, “These policies represent a starting point rather than a complete answer. Developers have the best understanding of the specific risks their products may pose.”
Sora Context Highlights Timing of Safety Focus
This recent focus on safety aligns with OpenAI’s recent updates to strengthen protections within its Sora AI video platform, including enhanced moderation for teen users, content filtering across video and audio outputs, and regulations regarding likeness and consent.
Shortly after these enhancements, OpenAI announced it would dissolve the Sora app and end its partnership with Disney, abruptly concluding one of its most prominent consumer AI initiatives.
While the company has not directly connected these decisions, the timing accentuates the importance of how AI products are developed, moderated, and scaled, especially concerning younger audiences.
