OpenAI Secures Teen Apps with Open Source Tools

OpenAI Introduces Tools to Secure Applications for Teens

OpenAI has recently unveiled a set of prompts aimed at helping developers make their applications safer for teenagers. This launch, announced on Tuesday, is part of an initiative to provide specific safety policies for young users. These prompts are designed to be used with OpenAI's open-weight safety model, known as gpt-oss-safeguard.

The goal is to simplify the process for developers looking to integrate safety measures into their applications. Instead of creating solutions from scratch, they can rely on these prompts to enhance the security of their creations. The areas covered include:

• graphic violence and sexual content
• harmful body ideals and behaviors
• dangerous activities and challenges
• romantic or violent role-playing
• age-restricted goods and services

These safety prompts are designed to be easily integrated with other models, although they are particularly optimized for the OpenAI ecosystem.

Collaboration with Security Experts

To develop these prompts, OpenAI collaborated with organizations specializing in AI safety, such as Common Sense Media and everyone.ai. This collaboration aims to ensure that the safety policies are relevant and effective.

Robbie Torney, head of AI and digital assessments at Common Sense Media, emphasized the importance of these policies. According to him, the fact that they are published as open source allows for adaptation and improvement over time, thereby enhancing safety across the ecosystem.

Ongoing Security Challenges

In a blog post, OpenAI acknowledged that even experienced teams face challenges in translating safety objectives into clear operational rules. This can lead to gaps in protection, inconsistent application, or overly broad filtering.

“Clear and well-defined policies are essential for effective security systems,” the company wrote. While these new policies do not solve all the complex challenges related to AI safety, they build on OpenAI's previous efforts, including product-level safety measures like parental controls and age prediction.

History and Controversies

Last year, OpenAI updated its guidelines for its large language models, called Model Spec, to better manage interactions with users under 18. However, the company has faced criticism regarding its safety track record.

OpenAI is facing several lawsuits filed by the families of individuals who committed suicide after excessive use of ChatGPT. These incidents highlight the potential dangers when users circumvent the built-in protections of the chatbot. While these new policies are not foolproof, they represent a significant step forward, particularly for independent developers seeking to create safer digital environments for young people.