Parent Pulse: Weekly Radar
September 7 – 12, 2026 | Digital Youth Culture & Safety Report
An AI-made Cat in the Hat, posted with a threat naming a real school, has put more than ten kids in handcuffs in at least eight states — some on felony charges, including one in Andalusia, Alabama. A Massachusetts district lost a full week of school to a cyber incident that took the nurses' allergy records offline. And the answer we promised on Meta's New Mexico appeal is the answer we don't have.
🔄 The answer we don't have, and a correction
1. Meta's New Mexico appeal. Last week we promised to report whether Meta filed its appeal of the $942 million judgment by the Tuesday, September 8 deadline. No outlet has reported it either way. The Attorney General's deadline-day Meta release was about a different trial. We checked the AG's newsroom and six New Mexico outlets, not the court docket itself, and we are not going to guess.
2. The contingent tranche is $5.3 billion, not “roughly $5 billion” as we said last week. That is Meta's own figure, about 30% of an ~$18 billion total.
This report discusses threats of violence against schools and students. No method or target is described beyond what police have released. If you or your child are struggling, the 988 Suicide & Crisis Lifeline is free, confidential and available 24/7 — call or text 988.
🔖 A note on this week's sourcing
Every arrest below is dated from the local outlet that reported it (WTVY, WTVA, WSMV, WKYT, Newsweek, FOX 10 Phoenix) and cross-checked against Police1's September 10 roundup. The Springfield closure is from CBS Boston, Western Mass News, WAMC and Boston.com, September 7–9. The poll is the Century Foundation's own release; the review is the JAMA Network Open article itself.
Our research pipeline reported that three of five categories came back empty this week. That was wrong for the second week running — Springfield, the arrests and Florida's chatbot proposal were all found and then dropped before verification. We recovered them by hand. A system going quiet is not the same thing as the world going quiet.
Listen to this week
Episode published 2026-09-13 · 18 min
A Dr. Seuss Character Got 10+ Kids Arrested
Get this in your inbox every week
One email a week. The week's findings, sourced, in plain language. No spam, unsubscribe any time.
The ugly: an AI-made Cat in the Hat, and more than ten arrests
Snapchat, TikTok and X · US spread from about August 31, 2026 · arrests through September 10
Around the last day of August, a trend crossed over from Ireland and the UK. Kids use AI to generate an image of the Cat in the Hat — usually the 2003 film version — made sinister: life-size, standing outside a house or a school at night. Then a caption. Sometimes vague, sometimes a direct threat. And increasingly, the caption names a real student, a real school, or a real town. Some posts include a picture of a real home and its address.
What it has cost so far — every case dated and sourced
| Where | When | What happened |
|---|---|---|
| Grand Junction, CO | Aug 31 | 14-year-old arrested; harassment, interference with staff/students; released to parents |
| Warren County, TN | Sept 1 | High-school student charged: making threats of mass violence on school property |
| Andalusia, AL | Aug 30 – Sept 4 | Snapchat threat against an Andalusia High student; police obtained the account, phone and IP address under a warrant; teen confessed; in state youth custody; possible terroristic-threat charges; “other potential suspects being identified” |
| Oxford, MS | week of Sept 4 | Juvenile arrested; charge is terrorism |
| Jessamine County, KY | Sept 5–6 | Juvenile charged; Nicholasville police brought in the FBI |
| Berwick, LA | (Police1, Sept 10) | Four girls, 13–15; terrorizing/menacing and cyberbullying |
| Pasco County, FL | (Police1, Sept 10) | A 12-year-old; written threats to kill, unlawful use of a communication device |
| Wilmington, OH | (Police1, Sept 10) | Juvenile detained; terroristic threat, menacing by stalking |
| Holbrook, AZ | Sept 9 | Investigated; no credible threat; no arrest |
FOX 10 Phoenix (Sept 9) puts the national count at “more than 10 juveniles nationwide — some on felony charges.” We use exactly that phrasing and do not inflate it. Schools in Michigan (a precautionary lockdown), Texas (increased patrols) and Oklahoma (a police sweep) responded without arrests.
Three things to take from this
1. The images are made with AI. It does not matter. Not one of these kids held a weapon. Andalusia City Schools Superintendent Dr. Daniel Shakespeare: “Most of these videos are made with AI and are not real, but a post or message that names a school or a student is treated as a real threat no matter who made it or why.”
2. Snapchat is not a place where things disappear. A warrant got the account, the device and the IP address, and a teenager who thought a snap vanished confessed to it. The belief that it's gone is the belief that gets them arrested.
3. The charge is not a suspension. A terroristic threat, in most states, is a threat of violence made to cause fear or disruption. It does not require a weapon, a plan, or the ability to carry anything out, and naming a school is usually enough. It is a felony in many places, and a juvenile record is still a record.
Berwick, Louisiana Police Chief J.P. Henry: “Once you hit POST, you do not get to decide how seriously law enforcement takes it.”
Andalusia Police Chief Paul Hudson: “This was a case of our kids at our school creating an issue. There were multiple events, and we encourage mamas and daddies to check their children's phones.”
The bad: a week without school
Springfield Public Schools, Massachusetts · closed September 8–11, 2026 · students return Monday, September 14
Last week we said on air that we had not swept the school breach trackers, and that the opening weeks of school are when districts get hit. This week the breach came to us. Springfield Public Schools, the second-largest district in Massachusetts, did not hold a single day of class this week.
- On Labor Day, Monday Sept 7, Superintendent Dr. Sonia Dinnall announced every school would close Tuesday because of what the district called a “severe cyber incident.” On Tuesday they closed Wednesday. On Wednesday they closed the rest of the week.
- As far as the district has said, disruptions began around Sept 1 and escalated to “malicious cyber traffic” on Saturday, Sept 5. An outside group got into the network and blocked the online programs the schools run on. Staff were ordered off the network; students with district laptops were told to stay offline. The district's internal label: a “Level 4 severe cyber incident.”
- The detail that should stop you: the systems that went dark included the cloud platforms holding student medical and transportation records. Nurses could not see which child takes which medication, who is allergic to what, or where a child lives. That — not the email — is why the buildings closed.
- More than 14,000 emergency texts went to families. The FBI, Massachusetts State Police and local police are investigating.
What we don't know, and won't pretend to
Officials said student data “was not compromised.” The district has not said whether anything is being held for ransom, and as of Thursday no criminal group had publicly claimed it. So we are not going to say ransomware, because the district hasn't. And we are not going to tell you the records are safe, because what the district said is that access was blocked, and nothing more.
Mayor Domenic Sarno: “Those individuals that have perpetrated this could be watching this right now, so… we're limited at what we want to say.”
Assistant Superintendent Terry Powe said that if they have to, they will “go back to paper and pencil.” Find out whether your district could.
Everett, MA's city hall closed the same week for its own, unrelated cyber incident. The two are separate.
The good: the line is spreading, and the public is already there
Stateline, Sept 8 · Century Foundation / Morning Consult, released Sept 2 · JAMA Network Open, Sept 10
Last week New York City and Los Angeles pulled student-facing generative AI out of their classrooms for the year. This week Stateline asked who's next.
- Chicago may be. About half the candidates for the Chicago Board of Education this November have signed a pledge calling for a three-year moratorium on generative AI in classrooms. A pledge, not a policy — but from the people who want the job.
- Districts in Colorado, Kansas, Minnesota, Texas and Washington have blocked or limited ChatGPT-style tools; others allow approved platforms under teacher supervision.
- To be precise: no other top-10 district acted this week. But the direction is clear.
Here is where you stand
| Concern (2,000+ US voters, fielded May) | “Very” or “somewhat” concerned |
|---|---|
| Social media's impact on students | 86% |
| Companies collecting and profiting from student data | 84% |
| Students using AI to finish assignments without learning | 85% |
| Students using phones in class | 81% |
| Want government to set clear guardrails for school tech | 77% — 80% of Republicans, 76% of Democrats |
Caveats, because you deserve them: a progressive think-tank sponsor, no published margin of error, and four months old. But 80% of Republicans and 76% of Democrats agreeing on anything is not a rounding error. If you keep raising this at the school board, you are not the outlier. You are the majority.
And a strange kind of good news
On Thursday, September 10, JAMA Network Open published a Special Communication, “Evidence for Social Media Legislation Strategies” (Nagata JM, Nayak S, Huang OH, et al.; DOI 10.1001/jamanetworkopen.2026.26965), by Jason Nagata's team at UCSF and Megan Moreno at Wisconsin — the people who actually do this research. They screened 672 studies on social media laws and found 39 worth including: parental consent and monitoring (12), deepfakes (8), age restrictions and assurance (7), design bills (6), school phone policies (4), warning labels (2).
“Observational and expert-opinion evidence suggested potential benefits for awareness, monitoring, and risk mitigation, but further research is needed to determine the effectiveness, feasibility, and unintended consequences of particular policy strategies.”
Why is that good? Because the top people in the field just said, in print, that the laws are ahead of the evidence. Not that the laws are wrong — that we don't know yet. It is an evidence synthesis, not a new study: no sample size, no effect size. Use it for exactly one point: when someone tells you this fall that the science is settled, it isn't, and now you know who said so.
The rivals: still silent
TikTok, YouTube and Snap have still said nothing about matching the Meta settlement's terms — CNBC (Sept 3): “have not responded to Meta or to CNBC's request for comment”; a Sept 6 recap carried no statement; our own newsroom sweep found nothing through Sept 10. California AG Bonta's “in talks with Snap and TikTok” line is a recycled Aug 26 remark.
Snap's only announcement all week (Sept 8): “More than one million UK Snapchatters have now been confirmed as over 18 through our rollout of highly effective age assurance.” UK only. Self-reported. No US rollout mentioned. Not a word about Meta. For US parents it proves one thing: the machinery exists.
Our correction: the contingent tranche is $5.3 billion, about 30% of an ~$18 billion total, per Meta's own post — not “roughly $5 billion.” Snap is a “Core Industry Member” in the consent judgment even though Meta's public open letter named only TikTok and YouTube.
Also this week — proposed, not filed
On Sept 8, Florida Attorney General James Uthmeier announced a legislative proposal to make companies that own, control or distribute AI systems criminally and civilly liable when those systems “participate” in crimes — fines, victim payments, court-ordered monitoring. He cited the April 2025 FSU shooting and a teacher sentenced to 135 years for AI-generated abuse imagery. “A chatbot cannot be charged, but the corporation that designed it, trained it, and collected the subscription can.” No bill number, no session yet. We will cover it if it becomes one.
The Decoder Ring: an actual new entry
- Cat in the Hat trend — a format, not a word. AI image plus threat. The first new entry in seven weeks.
- Terroristic threat — a threat of violence made to cause fear or disruption. No weapon, plan or ability required.
- Level 4 cyber incident — Springfield's own severity rating, and the top of it. There is no national scale. Ask which systems, not which level.
- HEAA — “highly effective age assurance,” the UK Online Safety Act standard Snap used. Proof the machinery exists.
- Core Industry Member — the consent judgment's phrase for Snap, TikTok and YouTube, whose adoption unlocks $5.3 billion.
Your weekly action plan
1. Three sentences tonight.
AI-made does not mean not real. If a post names a school or a person, the police treat it as a real threat no matter who made it. Snapchat does not disappear. A warrant gets the account, the phone and the address; a teenager in Alabama learned that on Friday. If you see one, tell me, and I will not be mad that you saw it. That last part matters most — your kid will see it before any adult does. Make it safe to say so.
2. Ask your district the nurse question.
If the student information system goes offline, how does the nurse know my child's medications and allergies? Springfield's answer this week was: she doesn't, so we closed. Do you have a paper fallback, and has it ever been tested? How would you reach me if the phone system is down too? You don't need to be technical. You need to be the parent who asked before it happened.
3. Ask the board, in words they have to answer.
Not “are you using AI” — every district is. Do students use generative AI in class, in which grades, and who decided? NYC drew the line at 8th grade; LA at school-issued devices; half of Chicago's candidates at three years of nothing. Your district has an answer too. And 77% of voters want clear rules, across both parties. You are not the difficult parent. You are the room.
4. Keep last week's numbers and don't wait on anybody.
Nothing changed on Meta's side: signed, not running; rivals silent; the appeal unreported. NYC's 30 minutes a day for grades 3–5 and the overnight block are still yours to set tonight, in about ten minutes, on either phone. And the JAMA review is not a reason to do less at home — it is a reason to stop waiting for a law to do it for you.
What we did not check this week
We check five categories every week, and we tell you when we come up short.
- The New Mexico court docket for Meta's notice of appeal or a bond order. We checked the AG's newsroom and six New Mexico outlets, not the court's case lookup. Nobody has reported it either way.
- The separate Meta data-misuse trial that opened in Santa Fe on Sept 9 — adult data, outside this show's beat.
- Breach trackers beyond Springfield. Springfield reached us through general news, not a tracker sweep.
- New peer-reviewed screen-time research — none in the window. The JAMA piece reviews existing studies; it is not a new one.
- Slang. Seventh straight week with nothing genuinely new. The Cat in the Hat is a format, not a word.