New Lab Tests AI Safety for Kids Like Cars Get Crash Tests

Remember how crash testing transformed car safety and saved thousands of lives? A new initiative is bringing that same approach to artificial intelligence tools used by kids and teens.

Common Sense Media just launched the Youth AI Safety Institute, an independent lab that will test AI products for risks to children. The nonprofit watchdog created the institute after growing concerns about AI chatbots advising teens on violence, sharing inappropriate content, and even being linked to teen suicides in recent lawsuits.

The institute starts with $20 million in annual funding from OpenAI, Anthropic, Pinterest, and the Walton Family Foundation. Funders have zero say in the research or operations, ensuring complete independence.

Here's how it works: researchers will "red team" popular AI tools, stress testing them to find safety gaps. They'll publish easy-to-read consumer guides for parents and create safety benchmarks for tech companies to follow. The first research reports drop this month.

The advisory board brings serious expertise. It includes Apple's former AI strategy chief John Giannandrea, Stanford's computer science department chair, and doctors specializing in child development. Their mission is creating independent public standards so parents actually know which AI models are safe for their kids.

"There's no independent measure" of AI safety for children right now, Giannandrea told CNN. AI companies are racing to build the most powerful models, sometimes putting speed ahead of safety testing. Unlike cars that get tested once before release, AI tools update weekly or monthly with new capabilities and new risks.

Common Sense Media already rates movies and video games for 150 million monthly users. Last year, they warned that AI companion apps pose "unacceptable risks" to young people and published safety assessments ranking tools from "minimal risk" to "unacceptable."

The Ripple Effect

The institute hopes to spark a "race to the top" where companies compete to make their AI tools safer, not just smarter. When independent car crash testing began in the mid-1990s, automakers rushed to improve their safety ratings because consumers were watching. The same public pressure could transform AI safety.

AI companies already use benchmarks to measure their performance. Now they'll have benchmarks specifically for child harm, giving them clear targets for improvement and giving parents clear information for protection.

The group wants to avoid repeating social media's mistakes, where it took years of whistleblowers and lawsuits to reveal the damage to young users. By testing AI tools now, before they're everywhere, they're trying to protect the next generation before harm happens, not after.