In a recent interview this past fall, former Surgeon General Vivek Murthy offered a trenchant analogy for how technology is affecting the experience of being a kid. Imagine giving your child a car with no seatbelts, brakes that don’t work so well, and letting them loose onto roads with no speed limits or traffic lights. That, said Murthy, is roughly what it is like to be a young person in 2026 trying to navigate the world of social media and AI amid a growing youth mental health crisis. Murthy pointed out that many kids turn to AI chatbots for emotional support and that the absence of safety checks, oversight, and meaningful guardrails only adds to parents’ growing panic that not enough is being done to rectify the problem.

All the more reason for Common Sense Media to do everything in its power to strengthen kids’ AI safety.

Last year, we helped enact a groundbreaking age assurance law in California and social media warning label laws in New York and California. In October, we proposed a ballot initiative called the California Kids AI Safety Act to build on that success in 2026 by preparing and protecting kids for the AI era. In December, OpenAI filed a competing initiative to block our effort.

We refused to let that stop us. When 80-90% of California voters – regardless of their party – demand stronger AI protections for our kids, we have a responsibility to get that done. So, rather than confuse voters with competing measures, we decided to work together to enact strong protections for kids, teens, and families.

At this pivotal moment for AI, America can’t make the same mistake we did with social media – when companies used our kids as guinea pigs and helped fuel a youth mental health crisis here in the U.S. and around the world.

So Common Sense wrote what will be the strongest youth AI safety measure in the nation’s history, and OpenAI agreed to replace its prior initiative with this one instead. As we have done in the past, we plan to enact these protections through the legislature or on the ballot. Kids win either way.

Here’s what the law will do to protect kids and empower parents:

First, it requires age assurance to protect AI users under 18 – and protect their privacy. If AI companies can’t be sure whether a user is a child or adult, they have to apply child-protecting settings.

Second, it prohibits child-targeted advertising and the sale of personal data for all California kids and teens.

Third, it requires comprehensive safeguards that protect kids and teens from the harms that worry parents most – AI generation and promotion of self-harm, eating disorders, violence, and sexually explicit acts, as well as manipulation of kids by creating emotional dependence, simulating romantic relationships, or making child users think they’re talking to a human.

Fourth, it requires AI companies to give parents powerful, easy-to-use parental controls to monitor and limit their kids’ AI use – and receive parental alerts if their kids show signs of self-harm. Parents can prevent the risks of prolonged exposure by turning on time limits and turning off the AI system’s memory.

Fifth, it requires AI systems to undergo independent audits of child safety risks. Common Sense has long provided parents with independent research and risk assessments of emerging platforms. This law will require annual risk assessments and independent audits, enforced by the California Attorney General. Child safety is too important to just take the companies’ word for it.

Finally, the law will give the Attorney General the power and authority to write the rules of the road and hold AI companies accountable through investigations and financial penalties for non-compliance and misconduct.

One leading editor calls the new initiative “a surprise move that could reshape the future of AI regulation for children and teens.” We’re determined to keep pushing for strong youth AI safety protections that can keep pace with the technology itself. Vivek Murthy is right: kids need seatbelts, brakes, speed limits, and traffic lights for what could be a wild ride ahead.