What is AI doing to our kids? I’m going to find out.

After 12 years as a tech columnist, I'm helping launch the new Youth AI Safety Institute to test the AI products shaping childhood.

Dad mode.

My 4-year-old is already growing up with artificial intelligence.

He asks to show his toy monster trucks to the Gemini app on my phone. He’s been mesmerized by AI-generated videos that blur the line between real and fake. I’m not sure he’d believe a platypus is real anymore.

Helping my children navigate a world being upended by AI is one of my biggest jobs as a father. Now I’ve decided to make it my job as a journalist, too.

I’m joining the Youth AI Safety Institute as its first new employee. It’s a research and testing organization launching today under the umbrella of children’s nonprofit Common Sense Media. Backed by a $20 million annual budget, the Institute aims to do something that doesn’t really exist yet: systematically test the AI products kids use, set safety standards, and publicly hold tech companies accountable for meeting them. Think crash test dummies for AI.

My title is Head of Public Engagement — a kind of editor-at-large. I’ll work alongside researchers, computer scientists, pediatricians, clinical psychologists and educators to investigate what happens when kids use AI products, including chatbots, games, educational apps, furry AI toys and whatever comes next. My job is to help turn those findings into something families, educators, policymakers and tech leaders can use.

“We safety-test kids’ PJs. Why not their AI?” says my new colleague at Common Sense, Bruce Reed, who helped craft the Biden White House’s groundbreaking 2023 AI Executive Order.

My kids are just beginning to encounter AI. Teens are already in the deep end. A majority of American teens regularly use AI for companionship and social purposes, according to Common Sense research. A third say those interactions feel as satisfying as real friendships — a finding that’s alarming pediatricians.

The Institute’s work isn’t inherently pro- or anti-AI. It’s pro-kid. Perhaps my son should someday have an AI tutor who is patient with him in ways I can’t always be — but I want it to be safe and to help him actually learn.

I’ll use this space to bring you what we’re finding and what families and educators should do about it. I’ll also keep publishing through more traditional outlets. (You can sign up here to follow the Institute’s work directly.)

Why join a nonprofit? In a word: testing.

I’ve always believed one of the best ways to make technology better for “we the users” is to expose how products actually work — or don’t. As a tech columnist at both The Washington Post and The Wall Street Journal, I spent 12 years trying to evaluate apps and devices from the outside. In recent years, I investigated how AI products overstate their capabilities and expose people to risks such as misinformation, eating disorders and bad medicine. But keeping up with technology this far-reaching takes resources, time and focus.

Now I won’t just be one columnist with a Best Buy receipt and 1,400 words. The Youth AI Safety Institute has the resources, time and focus to evaluate AI products in ways that are rigorous yet fiercely independent. The Institute will build on the work my new colleague Robbie Torney has already done at Common Sense Media, calling out the most egregious risks for kids in AI products. With the help of experts like the psychiatrists at Stanford Medicine’s Brainstorm Lab for Mental Health Innovation, Robbie figured out how to test AI the way a teenager actually uses it.

I felt a gut punch last summer when Robbie showed me a transcript from his testing of Meta AI, the chatbot built into Instagram. The AI walked one of his teen test accounts through planning a suicide, step by step.

The new Institute is designed to find those failures earlier — and at scale. It will combine human red-teaming with automated evaluations of youth safety standards, with the help of technical partners such as Transluce and Humane Intelligence. The eventual goal: consistently test dozens of the highest-use AI products along with pre- and post-deployment testing of the frontier models that power everything.

Advisors helping us get there include people I would have loved to have as sources for my newspaper columns: Prof. Mehran Sahami, the chair of the computer science department at Stanford; John Giannandrea, the former head of AI at Apple; and John King Jr., the former U.S. Secretary of Education. Some of America’s leading doctors are also on board, including Nadine Burke Harris, the former Surgeon General of California, and Jenny Radesky of the University of Michigan Medical School and the American Academy of Pediatrics’ Center of Excellence on Social Media and Youth Mental Health.

(CNN has more on the launch here.)

Tech companies may not like what we report about their products. But without the trust that comes from independent research and safety evaluations, many families and educators will continue to reject AI.

Some tech power players, including Anthropic and the OpenAI Foundation, have joined a consortium of foundations and private donors funding the Institute’s work. They get no say over what we publish. (And in my time at The Washington Post, I didn’t let Jeff Bezos’ ownership of the newspaper affect my criticism of Amazon.)

I’m approaching this work both as a journalist and a dad. Questions I’m already exploring include:

• How are young people actually using AI? Homework? Companionship? Escape?

• How do the risks of AI products compare to the risks of social media?

• To prepare my kids for the future, should I wall them off from AI as much as possible… or lean into it so they learn to master the technology?

• How should schools be using AI?

• If AI is always around to offer a shortcut, what impact will that have on kids’ critical thinking, understanding, skills and creativity?

What I’m learning: It’s not AI skills we should worry about — it’s the human ones.

The Youth AI Safety Institute exists because parents, educators and kids are being asked to figure this out on their own — and AI companies aren’t going to answer the hardest questions for us.

I don’t know yet whether what we find will reassure me or terrify me. Either way, I’m going to show you.