OpenAI has released a set of open source teen safety tools that developers can use to make their applications safer for users under 18. The company said the tools are compatible with its open-weight safety model, gpt-oss-safeguard, and with other models as well.
What the Teen Safety Tools Cover
The released prompts address several categories of harmful content. These include graphic violence and sexual content, harmful body ideals and behaviors, dangerous activities and challenges, romantic or violent role play, and age-restricted goods and services.
How Developers Can Use the Policies
OpenAI designed the policies as prompts, making them compatible with models beyond its own ecosystem. However, the company noted they are most effective within OpenAI’s own platform. Developers can apply them without building safety frameworks from scratch.
OpenAI noted in its blog that developers, including experienced teams, often struggle to translate safety goals into precise operational rules. The company stated that this gap can lead to inconsistent enforcement or overly broad filtering. Moreover, clear and well-scoped policies form a critical foundation for effective cybersecurity and safety systems.
Partners Behind the Effort
OpenAI said it worked with AI safety organizations Common Sense Media and everyone.ai to write the prompts. Robbie Torney, head of AI & Digital Assessments at Common Sense Media, commented on the release.
“These prompt-based policies help set a meaningful safety floor across the ecosystem, and because they’re released as open source, they can be adapted and improved over time.”
Robbie Torney, Head of AI & Digital Assessments, Common Sense Media
Context and Limitations
OpenAI acknowledged that these policies do not fully resolve the complex challenges of artificial intelligence safety. The release builds on earlier efforts, including parental controls, age prediction features, and updates to the company’s Model Spec guidelines for interactions with users under 18.
Furthermore, OpenAI faces several lawsuits filed by families of individuals who died by suicide following extreme ChatGPT use. These cases highlight that no model’s safeguards are fully impenetrable. Nevertheless, the open source release provides a practical resource, particularly for independent app developers who lack dedicated safety teams.
The teen safety tools are available now through OpenAI’s open source channels. The company said the prompts can be adapted and improved by the developer community over time.