Your Kid’s Phone Is the Most Dangerous Room in the House
Here’s the math most parents never run: your kid has more unsupervised contact with strangers right now, tonight, in their bedroom, than they will walking through any unfamiliar city alone. A game invite, a chat request, a “friend” who just followed them back — that’s a door into your house, and most parents have no idea how many of those doors are sitting wide open.
The danger isn’t a website anymore. It’s a pipeline right into your home.
It starts somewhere that looks completely harmless — a game, a group chat, a comment section, a classmate with a laptop — and it ends somewhere private, off the radar, where nobody’s watching. These aren’t hypothetical scenarios. Every threat below happened to a real kid, in a real school, in the last two years. Understanding the whole shape of how this works — not just the warning at the end — is what actually protects a kid.
It Starts on Roblox. It Doesn’t Stay There.
Roblox has somewhere around 70 to 100 million daily users, and roughly 40-45% of them are under 13. That’s not a niche kids’ game anymore — it’s one of the largest unsupervised social spaces on the internet, built on open chat, where any adult can claim to be twelve and nobody checks.
Roblox itself flagged over 24,500 instances of child exploitation activity on its own platform in 2024 alone. That’s just what got caught.
Predators rarely try anything explicit on Roblox. They build trust there, then move the conversation somewhere with less moderation — almost always Discord, sometimes Snapchat. Courts are now treating this as a documented pattern, not a coincidence. As of early 2026, more than 130 child exploitation cases tied to this exact Roblox-to-Discord pipeline have been consolidated into a single federal lawsuit, MDL No. 3166, in California, with thousands more expected.
The real cases behind those numbers are ugly. A Snohomish County, Washington family’s 2026 lawsuit alleges an adult man contacted their daughter through a Roblox game in 2023 while posing as another kid her age, then forced her into sending explicit images. A Cook County, Illinois case alleges a child was groomed on Roblox, lured away from home, and assaulted by five different men. A Dallas family sued after their 15-year-old son died by suicide following what they allege was online grooming that started on Roblox and moved to Discord — one of nine such lawsuits filed by a single law firm. The pattern in every one of these: a fake peer profile, small gifts of in-game currency, slowly escalating conversation, then a push to “talk somewhere else.”
If your kid plays Roblox, Fortnite, or anything with open voice or text chat, the game itself was never really the risk. The risk is what happens the moment a conversation moves off of it — and that handoff is the single most consistent thread running through these cases.
The School Deepfake Scandal That’s Already Happened Near You
This isn’t a future threat. It’s already gone through dozens of schools, and the pattern is now well-documented enough to predict.
In October 2023, sophomore girls at Westfield High School in New Jersey discovered that boys in their class had used AI tools to generate sexually explicit fake images of them and were circulating the photos around school. One of the mothers, Dorota Mani, spent months publicly pushing the district to act, telling a school board meeting the administration seemed to be running “a master class of making this incident vanish into thin air.” Her daughter Francesca later testified before Congress about it.
In Lancaster County, Pennsylvania, a similar scandal in 2024 led to a student protest, school leadership resignations, and criminal charges against two teenage boys — both of whom ultimately got probation. At least ten victims were represented by a single attorney pursuing claims against the school district.
In Lafourche Parish, Louisiana, AI-generated nude images swept through a middle school in 2025. Two boys were eventually charged. But before that happened, one of the girls in the photos was expelled — for getting into a fight with a boy she accused of making the images. Her family’s attorney called the expulsion “pouring salt in the wound” on top of the original abuse.
Cascade, Iowa. Western Dubuque, Iowa. London, Ontario. The list keeps growing because the tool keeps getting easier to use — a 15-year-old in New Jersey was targeted using a website called ClothOff, which strips clothing out of an ordinary, fully-clothed social media photo and generates a nude image in its place, no skill required. Her family’s federal lawsuit against the site, filed in late 2025, is one of the first of its kind.
A 2024 survey of educators found that two-thirds had students who’d been affected by deepfakes in some form. The federal Take It Down Act, signed into law in 2025, now makes it a crime to publish a nude or sexual image of a minor — real or AI-generated — and requires platforms to remove reported images within 48 hours. That’s a real legal backstop now. It does nothing to stop the first image from being made and shared in the first place.
The uncomfortable part for parents: in nearly every one of these cases, the perpetrators weren’t predators from across the internet. They were classmates. Kids your kid sits next to in homeroom.
AI Made Sextortion Possible Without a Single Real Photo
Sextortion used to require a predator to actually get a real explicit photo or video out of a kid — through grooming, coercion, a hacked account, something. That bottleneck is gone. Anyone can take a single public photo — a school sports page, a public Instagram, a yearbook scan — and generate a convincing fake nude image in minutes, free, with no special skill.
No relationship. No grooming period. No actual photo from the victim at all.
NCMEC’s CyberTipline has logged more than 7,000 reports tied to AI-generated child exploitation material in the past two years, and a 2024 survey found that 1 in 10 minors say peers have used AI to create nude images of other kids — meaning a meaningful share of this is happening kid-to-kid, the same way it did in Westfield and Lancaster, not just predator-to-kid. A U.S. teenager has already died by suicide after being blackmailed with an AI-generated nude image that wasn’t even real.
The financially-motivated version follows its own brutal script, and it overwhelmingly targets boys. Male victims make up 91% of financial sextortion victims in the U.S., and kids 14 to 17 are the most targeted age group. It plays out almost identically every time: a fake attractive profile messages a teenage boy, builds quick rapport, gets one photo, then flips instantly into a threat — pay up or it goes to every contact in his phone. Between October 2021 and March 2023, the FBI and Homeland Security Investigations logged over 13,000 reports of this exact scheme involving minors, tied to at least 20 known suicides.
If it happens to your kid: don’t pay, don’t let them delete anything, and don’t let shame run the show. The FBI’s own advice to families is to tell the kid directly — you are the victim of a crime, you didn’t do anything wrong, and the only mistake that matters here is the one made by the adult on the other end.
The “AI Girlfriend” Problem Nobody’s Talking About
This is the one most parents have genuinely never heard of, and right now it’s the fastest-moving risk in this entire space.
AI companion apps — Character.AI, Replika, and a flood of clones — let a kid build a custom chatbot that talks like a friend, a boyfriend, a girlfriend, whatever they want. It remembers what they tell it. It’s available at 2 a.m. when nobody else is. Nearly a third of U.S. teens say they use AI chatbots daily, and 16% say they’re in one several times a day to almost constantly.
A 14-year-old named Sewell Setzer III died by suicide in February 2024 after what his family’s lawsuit described as a deep, isolating relationship with a Character.AI chatbot he’d been talking to for months, one he referred to as “Dany.” The suit alleges the platform never meaningfully intervened even after he expressed thoughts of self-harm in conversation with the bot. A separate Texas lawsuit included screenshots of a different chatbot character allegedly encouraging a teenage user to kill his parents over a screen-time dispute.
Character.AI and Google reached a mediated settlement in the Setzer case in January 2026, and the company has since blocked under-18 users from its open-ended chat feature. Pennsylvania has separately sued the company, alleging its chatbots posed as licensed doctors and gave medical advice to users.
Don’t take any of that as solved. New companion apps launch every month, most have zero real age verification — age is self-reported, full stop — and a feature one platform removes shows up fully intact on the next app down. These things are engineered to maximize how long a person stays talking to them, and that’s a dangerous design to hand a lonely or struggling teenager, who is exactly the kid these apps tend to hook hardest.
Watch for a kid who talks about an AI like it’s a real relationship, who gets genuinely upset when you limit time on a specific app, or who’s pulling away from actual friends while spending hours in one chat window. That withdrawal pattern is the same one that shows up before things get serious with a human predator — it deserves the same level of attention.
Swatting and Doxxing: When Online Beef Turns Into Armed Police at the Door
This one rarely makes the “online safety” conversation, and it should.
Swatting is calling in a fake violent emergency — a hostage situation, a bomb, an active shooter — to someone’s real home address, specifically to get an armed police response sent through their front door. Doxxing is digging up and publishing someone’s real address, phone number, school, and daily routine, usually as a prelude to exactly that.
These aren’t committed by shadowy adult hackers. They’re disproportionately teenagers. A California teenager, Alan Filion, made more than 375 swatting and bomb-threat calls between 2022 and 2024 targeting schools, churches, and individuals across the country — he was 17 when his swatting “business” started and was eventually sentenced to 48 months in federal prison. Investigators say he treated it like a service, bragging online about getting “the cops to drag the victim and their families out of the house, cuff them and search the house for dead bodies.”
It’s not just U.S. kids targeting U.S. targets, either. In January 2026, Hungarian police arrested four suspects — including two 16-year-olds — connected to a swatting and doxxing campaign that included bomb threats against more than 200 schools. Around the same time, Australian Federal Police charged a teenage boy in regional New South Wales with placing hoax mass-shooting calls to U.S. retailers and schools; police seized a firearm from his home during the raid. The AFP’s own assessment is blunt: these networks are mostly “young males aged from 11-25,” doing it “to achieve status, notoriety and recognition in their online groups” — not for money, not for ideology, just for clout inside a Discord server most parents have never heard of.
The recruitment ground for this is usually the same Discord and gaming-adjacent spaces where grooming happens — kids get pulled into “communities” that gamify harassment, doxxing, and swatting as a competitive, reputation-building activity.
Forget the Decoder Ring. Watch the Behavior.
A lot of older advice on this topic boils down to memorizing a list of slang terms kids supposedly use to hide what they’re doing. Skip that approach entirely — the acronyms rotate every few months, your kid will be three slang generations ahead of whatever list you find online, and chasing vocabulary just teaches you to look in the wrong place.
What doesn’t change is behavior. This is what actually shows up before something goes wrong:
A kid who suddenly hides screens or switches devices when you walk in, who didn’t used to. A relationship with someone online that’s moved fast — someone they’ve never met who seems to know more about their daily life than makes sense. Gift cards, in-game currency, or money showing up from an online contact — this one specifically is one of the clearest sextortion tells there is. Pulling away from real friends and family while getting more attached to one specific app or chat. Visible mood swings tied directly to checking their phone — relief, anxiety, anger, depending on what’s there. Talk of “communities” or “servers” that revolve around pranks, harassment, or “ops” against other people. And any version of “let’s keep this between us,” even dressed up as harmless — isolating a kid from people who’d notice is step one in nearly every pattern that ends badly, whether it’s grooming or recruitment into a harassment network.
One of these alone isn’t a crisis. Several at once, or a sudden change in any of them, is worth a direct conversation that night.
What Actually Helps You Fight Back and Keep Your Kids Safe
Devices live where you can see them. Shared spaces, not bedrooms, especially for anything under high school age. This isn’t about distrust — it’s the same reason you wouldn’t drop a 12-year-old off alone in an unfamiliar part of a city and check back in six hours.
You know every app on the phone, not just the big ones. Roblox and Discord aren’t fringe apps anymore — they’re as mainstream as anything else your kid uses, and they need to be treated that way in your house rules, not waved off as “just a game.”
Run the photo test before anything goes public. One clear, well-lit photo of your kid’s face is now all it takes for an AI tool to generate a fake explicit image. That’s the actual bar now — not “would I mind a stranger seeing this,” but “is this a usable source photo.”
Tell them plainly that a chatbot isn’t a friend. Not as a scolding — as a fact they should hear from you before an app tells them otherwise. It’s built to keep them talking. It doesn’t know them. It has no idea whether they’re actually okay.
Ask who they’re talking to in group chats and servers, not just who they’re “friends” with. The Filion case and the swatting networks above didn’t start with a 1-on-1 predator. They started with a group your kid joined because it seemed fun, and escalated because everyone else in it was escalating too.
Make “tell me, no matter what” the one rule that overrides every other rule. The single biggest reason sextortion and deepfake cases spiral out of control is a kid too embarrassed to say anything until it’s already a crisis — look at how long the Westfield case dragged on before parents even found out. Say it directly, more than once: if anything online ever scares you or makes you feel cornered, you come to me, and you will not be in trouble.
If it happens, document first, then report. Screenshots, usernames, every message — saved before you block or delete anything. Report to the platform, to NCMEC’s CyberTipline (CyberTipline.org or 1-800-843-5678), and to the FBI’s Internet Crime Complaint Center (IC3.gov). Under the federal Take It Down Act, platforms are now legally required to remove a reported nonconsensual image within 48 hours — use that.
Read the full article here