California’s Adam’s Law, a bill that would police how companion chatbots treat children, has picked up a powerful backer: OpenAI, which wants Governor Gavin Newsom to sign it.
Ann O’Leary, OpenAI’s vice president of global policy, announced the support, saying California can set a strong standard for youth AI safety while federal action stalls. The bill, introduced in February, would require operators to assess and mitigate risks of physical, financial, and psychological harm to minors.
Under the amended text, chatbots aimed at children would need age determination, documented risk assessments, and published child safety policies. Default settings would limit sessions to one hour, cap daily use at two hours, disable persistent memory and push notifications, and ban custom system prompts. Chatbots would be barred from claiming sentience, expressing romantic interest, or diagnosing health conditions.
Independent safety audits would start in 2029, with civil penalties up to $15,000 per affected child for intentional violations, while most requirements take effect on July 1, 2027.
OpenAI says its own ChatGPT for Teens already ships with safeguards that cannot be turned off, including Quiet Hours and study reminders, and points to data showing nearly nine in ten teen users turn to the assistant for learning.
