OpenAI announced Tuesday that it will release a set of prompts that developers can use to make their apps safer for teens. AI Lab said its set of teen safety policies can be used in an open weight safety model known as gpt-oss-safeguard.
Rather than starting from scratch to find ways to make AI safer for teens, developers can use these prompts to enhance what they build. These address issues such as graphic violence and sexual content, harmful body ideals and behaviors, dangerous activities and tasks, romantic or violent role-play, and age-restricted products and services.
These safety policies are designed as prompts, so they are easily compatible with other models besides gpt-oss-safeguard, but are probably most effective within OpenAI’s own ecosystem.
To create these prompts, OpenAI said it worked with AI safety watchdogs, Common Sense Media, and Everyone.ai.
“These prompt-based policies help set meaningful safety bounds across the ecosystem, and because they are released as open source, they can be adapted and improved over time,” Robbie Torney, head of AI and digital assessment at Common Sense Media, said in a statement.
OpenAI noted in a blog that developers, including experienced teams, often struggle to translate safety goals into precise operational rules.
“This may result in gaps in protection, inconsistent enforcement, or overly broad filtering,” the company wrote. “Clear and broad-based policies are a critical foundation of an effective safety system.”
tech crunch event
San Francisco, California
|
October 13-15, 2026
OpenAI acknowledges that these policies are not a solution to the complex challenges of AI safety. However, this builds on previous efforts such as product-level safety measures such as parental controls and age prediction. Last year, OpenAI updated its guidelines for large-scale language models, known as model specs, to address how AI models should behave for users under 18.
However, OpenAI itself does not have a very clear track record. The company is facing several lawsuits brought by families of people who died by suicide after extreme use of ChatGPT. These dangerous relationships often form after users ignore the chatbot’s safeguards, and no model’s guardrails are entirely impenetrable. Still, these policies are at least a step forward, especially since they can help indie developers.
