OpenAI Backs California Law to Make Teen Chatbots Safer

OpenAI is urging California's governor to sign a bill that would turn voluntary teen-safety rules into legal requirements. Senate Bill 1119 targets AI companion chatbots used by minors. Under the bill, companies would have to figure out how old users are—or apply child protections by default. They would also need to run safety checks before launching new chatbots, publish child-safety policies, and pass independent audits.

Why does this matter to you? If you have kids or teens, they're likely using AI for homework, research, or just chatting. Nine in ten teens use ChatGPT for learning each week, according to OpenAI. But there are real risks. Chatbots can sometimes encourage self-harm, pretend to be human, show romantic interest, or create emotional dependence. This bill would force companies to limit how long teens can chat, turn off memory features by default, and require crisis-response tools when a teen shows signs of distress.

OpenAI also argues that AI should not be treated like social media. The bill makes that distinction, focusing on companion chatbots rather than feeds or autoplay. Parents would get tools to manage or shut off access for kids under 16. That's a big shift from today, where many safety measures depend on each company's goodwill.

The catch: formal rules and audits could make it more expensive to build AI products for minors, especially for smaller startups. And legislation often moves slower than AI evolves. Still, supporters say a clear legal baseline is better than trusting companies to police themselves. The bill is expected to take effect in July, if signed.