California: OpenAI Supports SB 1119 on AI and Minors

OpenAI highlights the built-in protections already integrated into ChatGPT for teenagers, such as parental controls, age prediction, and interaction rules. The company supports California's SB 1119 bill, which requires age verification, independent audits, and limits on targeted advertising, and calls on Governor Newsom to sign it. It also claims that nearly 90% of teenage users of ChatGPT use it to learn or gain independence.

OpenAI details existing safeguards for those under 18

OpenAI offers a version of ChatGPT specifically designed for teenagers, aimed at helping those under 18 learn, develop critical thinking, and create, with integrated protections and additional parental controls. If the system determines that a user is under 18, or if the user states an age between 13 and 17, access automatically switches to this version, where the protections cannot be disabled. The company emphasizes learning features such as quizzes, visualizations, and a study mode, developed with educators and experts, intended to guide students step by step and encourage stronger learning habits. Utilizing the context of conversations, including memory, helps to apply protections more consistently, identify concerning patterns, and support long-term learning by retaining useful information and preferences for ongoing projects. According to OpenAI, this continuity can also prevent the reinforcement of harmful patterns and facilitate more effective responses when help resources are needed. Over the past year, the company has added parental controls, age prediction, a Teen Safety Blueprint, and Principles for those under 18 to its Model Spec, presented as the set of rules guiding the behavior of its models. These rules specifically prohibit romantic engagements, encouragement of emotional dependency, and any claims that ChatGPT is human or sentient. Parental controls also allow for managing key settings and receiving notifications in high-risk situations.

SB 1119 establishes age verification, audits, and automatic protections

OpenAI presents SB 1119 as a text that combines strong protections with maintaining access to useful AI tools. The bill mandates determining users' ages, identifying and addressing security risks before making tools available to young people, and undergoing independent audits. It includes safeguards against harmful content such as self-harm and sexually exploitative material, the provision of parental guidance and limitation tools, connections to support resources in cases of serious risks, as well as limits on targeted advertising and protections for minors' personal data. For users aged 13 to 17, OpenAI asserts that these protections should apply automatically. The company states that the law would combine significant obligations and effective measures to ensure the safe use of AI by young people while recognizing that AI is not a social network. OpenAI notes that policymakers aimed to preserve access to educational and essential safety features, including responsible practices around ChatGPT's memory, and to promote age-appropriate experiences.

Letter to Newsom and call for the bill's signature

OpenAI wrote to Governor Gavin Newsom to commend his actions, as well as those of Senator Steve Padilla, Assemblymembers Buffy Wicks and Rebecca Bauer-Kahan, and the state legislature, and to encourage him to sign SB 1119. The company claims to support the bill, linking it to its previous initiatives such as its global policy principles, advocacy efforts in California, and involvement in the Parents & Kids Safe AI Act. It describes the bill as laying the groundwork for a robust framework for protecting youth in California and believes that continued research and cooperation during implementation could optimize key points and ensure the framework's relevance as technology evolves. OpenAI states it is willing to support these efforts and continue developing AI experiences for teenagers, incorporating enhanced automatic protections and learning opportunities. It characterizes SB 1119 as providing significant protections while preserving access to useful tools and emphasizes that in the absence of federal action, California can set a strong standard for youth safety in the face of AI.

Usage context: a majority of teens use it to learn

OpenAI asserts that nearly 90% of teenage users of ChatGPT use it weekly to learn, gather information, develop skills, or increase their productivity. The company advocates that experiences designed for young people should meet specific developmental needs, with strong protections by default and appropriate opportunities for learning, creation, and exploration. It argues that AI can benefit teenagers without compromising their safety and presents this goal as central to SB 1119.