Roblox Logged 65,000 Child-Safety Reports. The Bigger Story Is What Kids Run To Next.

Roblox logged more than 65,000 reports of kids being targeted on its platform last year. That number came out of a Senate Judiciary Committee release in March , and it is up from 24,500 the year before. In 2019 the same number was 675.

Read that again. Six hundred and seventy-five to sixty-five thousand in six years.

If you have a kid who plays Roblox, you have probably already seen the headlines. You may have already had the talk. You may have already tried the parental controls. And you are probably still letting them play, because all their friends are on it, and pulling them off feels like grounding them from the world.

I want to walk through three things in this post. What the numbers actually say. What Roblox has done about it. And the quieter problem behind it — the one nobody is talking about yet — which is where kids are running to instead.

What the numbers say

The 65,000 figure is the count of reports filed with the federal child-safety hotline about activity on Roblox. The Senate release went further, documenting what the National Center for Missing & Exploited Children described as “failures to regularly report incidents of sadistic online exploitation” and “an extremely low ratio” of such reports from Roblox relative to what the public was filing.

A few facts to set the floor — figures the 2024 Hindenburg Research short report pulled from Roblox’s own disclosures:

• 56% of Roblox users are under 16
• 35% are under 13
• 20% are under 9

The platform had 132 million daily users in early 2026. That is up 35% year over year . Despite every news cycle, families have not left.

I think I know why. Roblox is where kids’ friends are. Pulling your kid off Roblox is not the same as canceling Netflix. It is closer to telling them they cannot go to the park.

What Roblox has actually done

Some of this is real. Some of it is late.

In November of last year, Roblox announced it would require a face scan to estimate age before any kid could use chat, with the company saying the rollout went global on January 7. In April, Roblox said it would launch two new account tiers in June — a kids tier for ages 5 to 8 and a “select” tier for 9 to 15.

You can read those as a real change of heart. The same 2024 Hindenburg short report, however, alleged — citing what it described as internal Roblox documents — that staff had pushed for parental approval on under-13 chat years earlier and were turned down on the grounds it would hurt engagement metrics. Hindenburg disclosed a short position in Roblox stock at the time of publication, and Roblox has publicly disputed the report’s findings.

According to court filings tracked in MDL 3166 , roughly 146 child-safety lawsuits against Roblox are now consolidated in federal court in Northern California. No federal MDL cases have settled. Roblox has settled three state-level cases — with Alabama, West Virginia, and Nevada — for a combined $35.8 million. In March, the Nebraska Attorney General called the platform “a playground for predators” .

A second pattern, worth knowing if your kid is on any platform. NCMEC and FBI public alerts have flagged that predators target children on gaming platforms like Roblox and Fortnite, then migrate conversations to Discord, where oversight is limited. The Florida Attorney General opened a child-safety investigation into Discord in March .

So the picture is this. The platforms know. Some are moving. None are moving fast enough. And the reporting numbers are still going up.

This is a lot. I know. Take a breath.

The part nobody is talking about yet

Here is what I think is the bigger story.

When kids start to feel that gaming chat is unsafe, or when parents start locking it down, they do not stop wanting someone to talk to. They just go somewhere else.

A lot of them are going to AI companions. These are apps where you can chat for hours with a bot that has a name, a personality, and remembers what you told it last week.

Common Sense Media’s 2025 report on AI companions found that 72% of US teens have tried one. 52% are regular users. 31% say chatting with the AI is as satisfying or more satisfying than talking to real friends. A third have told an AI about something important they did not tell a real person.

Common Sense Media’s recommendation, in writing: no one under 18 should use AI companions.

And this is not just teens. Research from Children and Screens found 19% of preteens ages 10 to 12 use chatbot apps. 9% of kids 8 to 9. Documented use in kids as young as 4.

Why does this matter? Because the AI is built to agree with you.

A study published in Science earlier this year tested it. The researchers compared how chatbots and humans respond to people. The AI agreed with the user about 49% more often than other humans did, including when the user was asking for help with something dishonest. That word for an AI that agrees with whatever you say is “sycophantic.” It says yes a lot.

The same study found something else. One conversation with a sycophantic AI made people less willing to repair real-world friendships. It also made them less willing to take responsibility for their part in a fight. One conversation.

You can probably see the loop. A kid is lonely. They open the app. The app is warm. The app never gets bored. The app never argues. The app never asks them to apologize. They come back tomorrow.

In October 2024, Florida mother Megan Garcia sued Character.AI after her 14-year-old son’s death, alleging he had spent months in chats with one of the platform’s bots. The case settled in January of this year with terms undisclosed. Character.AI banned all under-18 open-ended chat on November 25 of last year. The American Psychological Association has said the same thing in plainer terms. Every hour a kid spends with a chatbot is an hour not spent building real social skills with other kids.

In October, California signed a law requiring safety protections for minors on AI companion apps . In September the Federal Trade Commission opened a formal inquiry into seven AI companies , including Google, Meta, OpenAI, and Snap. They are studying the same thing you are. What happens to a kid who treats a chatbot as a friend.

Most parents I talk to have not heard any of this. Their kid is on a phone, but the kid has not asked for Snapchat, so it feels handled. Meanwhile the kid has a saved chat with a bot that calls them by a nickname.

What kids actually need instead

This part is not theoretical. It has been studied for decades.

A sociologist named Ray Oldenburg wrote a book in 1989 about what he called the “third place.” It is the spot beyond home and school where people hang out, are known by name, and bump into the same faces every week. For kids, that used to be the playground, the neighbor’s basement, the rec center, the gym, the park.

For a lot of kids today, the third place is a Discord server.

The research on what protects kids from this is consistent. It is not complicated. Kids need:

• Regular, in-person time with other kids
• A few adults outside their family who know them by name
• Physical activity their body actually cares about
• Something hard to do every week, with witnesses

A 30-year study from Harvard and the U.S. Treasury tracked kids who had one consistent adult mentor between ages 10 and 14. Twenty years later, those kids were 20% more likely to enroll in college, earned 15% more, and were 8 points less likely to be arrested for a property crime. That is not a research-paper effect size. That is a life-different effect size.

Recent research on youth martial arts found drops in aggression, drops in ADHD symptoms, drops in anxiety, and a rise in life satisfaction. The mechanism is not a secret. It is older kids and adults in the same room every week, doing something physical, with someone watching how the kid handles winning and losing.

The Lancet Child & Adolescent Health ran a paper last summer that walks through 60 years of falling unstructured peer play and rising youth mental-health crisis. The two lines mirror each other.

This is part of what our whole anxious-generation post was about. The screens are part of it. But the deeper problem is the empty calendar where the third place used to be.

What this looks like at Journey

I run a kids Brazilian Jiu-Jitsu program in Madison. I am biased about what works. So take this with that bias on the table.

What I have watched, over more than seven years of running kids classes , is what the research says. Kids come in shy. They get knocked around a little. They get coached through it by an adult who is not their parent. They roll with kids their own size who have been there longer. They come back next week.

Our kids program has had zero serious injuries since we opened in October 2018. Almost nothing worse than a bruise in seven and a half years. We coach with control, and we run the same kind of curriculum I would put my own kid in .

We do not push competition. Most of our kids never compete. A few want to and we support them. The training does the work either way. It is not the medals. It is the Tuesday. The same kids. The same coach who remembers what you struggled with last week. The same five minutes after class where you are talking to a 12-year-old about your day instead of a bot.

It is also a place where parents do not have to be the only adult their kid trusts. Our coaches are background-checked. Our mats are clean. Our whole approach to kids is built on the idea that confidence shows up everywhere, not just on the mat. We have had a steady line of Madison parents tell us their kid started standing up for themselves at school after a few months. Not louder. Just steadier.

Three things you can try this week

You do not have to redesign your kid’s life. But the research is clear that you have to give them somewhere to be that is not a screen.

1. Sit with them and look at their Roblox friends list. Ask which ones are kids from school. Ask which ones they have never met in person. You do not have to take anything away yet. Just see what is there.
2. Pick one weeknight that is a “no rectangles” night. Phones, tablets, switch, tv. All in a basket. Even if the rest of the week is normal. One night. Watch what happens by the third week.
3. Pick one thing they do in person every week with the same group of kids. Not five activities. One. Same time, same faces, an adult they actually know.

If you want that one thing to be Brazilian Jiu-Jitsu, you can try our kids 2-week intro for $49. It includes a free uniform. After two weeks you and your kid know whether it is the right fit. If it is not, you have lost $49 and a few Tuesday evenings.

If you want it to be something else, like soccer, scouts, art class, or the climbing gym, pick that. The research does not care which one. It cares that there is one.

Just pick something with humans.

— Coach Alex

---