Download PDF
< View all Policies

Testing OpenAI Parental Controls Parental Controls: An Evaluation of Safety, Gaps, and Policy Implications

avatar Lyonne Zhu  November 20, 2025

This report evaluates OpenAI’s new parental controls for ChatGPT and examines how well they protect minors in real-world use. Through a series of behavioral tests with a simulated teen account, it identifies where safeguards work as intended and where significant gaps remain. The findings highlight concerns around age assurance, delayed enforcement, risky content exposure, and in-chat purchases.

The report outlines key improvements needed to ensure AI tools are truly safe and accountable for young users. It offers guidance for parents, policymakers, and industry leaders on what meaningful protection should look like as generative AI becomes a larger part of teens’ everyday lives.

Download PDF

Lyonne Zhu

Lyonne Zhu is the Digital Safety Tech Policy Fellow at the Family Online Safety Institute (FOSI). She is a second-year Master of Arts in International Relations candidate at Johns Hopkins University’s School of Advanced International Studies, where she focuses on technology policy, climate resilience, and sustainable development. Lyonne brings experience in policy analysis, digital communication, and program design from her work with city governments, international organizations, and nonprofits. At FOSI, she is passionate about making emerging technologies more accessible and ensuring that online spaces are safe for children and families.