OpenAI is backing California Senate Bill 1119, a proposed framework for youth AI safety.
The company published its support on August 31 and urged Governor Gavin Newsom to sign the bill. OpenAI’s argument is that young people should keep access to useful AI tools, but with safeguards that apply automatically when a product is used by teens.
That framing matters. OpenAI is not treating teen safety as a settings page that parents may or may not configure. It is describing age-appropriate AI as a default product layer.
The bill OpenAI wants signed
OpenAI says SB 1119 would require AI services to determine a user’s age, identify and address safety risks before making products available to young people, undergo independent audits, and protect young users from harmful content.
The company also points to requirements for parental tools, crisis-support resources in serious safety situations, limits on targeted advertising, and protection for young people’s personal information.
For users identified as ages 13 through 17, OpenAI says those protections should apply automatically.
The company also argues that AI is not social media and should not be regulated as a simple copy of social-platform rules. Its preferred framework preserves access to educational and safety-critical features, including responsible uses of memory, while still adding accountability and age-appropriate protections.
ChatGPT for Teens is the product example
OpenAI ties the policy position to ChatGPT for Teens, its under-18 experience. The company says users are automatically placed into ChatGPT for Teens when its systems estimate they are under 18 or when they state their age is 13 to 17.
OpenAI says the teen product includes learning-oriented features such as Quizzes, Learning Visualizations, and Study mode. It also says teen safeguards are part of the baseline experience, not optional controls that users can turn off.
The company is also defending the role of memory and other context features for teen users. Its argument is that context from prior conversations can help safeguards apply more consistently, support longer learning projects, and improve responses when a young person may need help.
That is likely to be one of the harder implementation questions. Memory can support continuity and safety detection, but it also raises data-handling, parental-visibility, retention, and consent questions. SB 1119’s real test will be whether implementation can preserve educational usefulness without making teen data collection too broad.
Youth AI safety is becoming a state-level policy race
OpenAI’s post is also a sign of where AI regulation is moving in the United States. In the absence of a broad federal AI law, states are setting more of the practical product rules.
California is especially important because a state-level standard there can become a national product baseline. AI companies often prefer one consistent product architecture to a patchwork of state-specific experiences, especially when the rules touch identity, age prediction, default safeguards, advertising limits, audit evidence, and crisis routing.
For OpenAI, supporting the bill gives the company a chance to shape the category before stricter or less product-aware approaches take hold.
