Parent Pulse: Weekly Radar
September 21 – 26, 2026 | Digital Youth Culture & Safety Report
A middle schooler used AI to make nude images of two classmates, and Kentucky charged him with four crimes. There was no hacker and no stranger — a free tool, a phone, and two kids who sit in the same hallway. Plus: Discord quietly decided how old your child is, a second district in three weeks lost its grade book, and a government sued OpenAI over a school shooting.
This report discusses AI-generated sexual images of children, a school shooting and a suicide. 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
Henderson is from FOX19 / 14 News (September 21), corroborated by The Messenger and The Hendersonian. Spokane is from the Spokesman-Review (September 22 and 24), KXLY and Security Magazine. Discord is from its own blog post of September 22. British Columbia is from CBC, Al Jazeera and Quartz. The phone-ban study is Frontiers in Developmental Psychology, 22 September 2026, with the date confirmed at Crossref rather than taken from the publisher page.
Three precision notes. No Spokane official has said “ransomware,” named a group or confirmed a demand, and the district's staff-only scope finding is explicitly preliminary. British Columbia's account of OpenAI's internal review is a government's allegation, untested in court. And four of our five Decoder Ring definitions are our own plain-language wording, not the dictionary's — Merriam-Webster's site blocks automated lookup, so we could not retrieve the verbatim text.
On process: for a third straight week our research pipeline dropped real stories before verification, and this time its summary declared four of ten threads “entirely empty.” Two of those contained this week's best reporting. We found them by reading the raw evidence rather than the summary. A pipeline that tells you a thread came back empty and a thread nobody searched are not the same thing.
Listen to this week
Episode published 2026-09-28 · 27 min
Middle Schooler Made AI Nudes of Classmates. Kentucky Charged Him
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: a classmate made the images
South Middle School, Henderson, Kentucky · police report Thursday, September 17 · first reported Monday, September 21
On Thursday, September 17, the Henderson Police Department received a report that nude photographs of middle school students were being generated with artificial intelligence and shown to other students. During the investigation, officers identified AI-generated photographs of two minor victims on a digital device.
A juvenile was detained and lodged at a juvenile detention center on four counts:
| Count | Charge |
|---|---|
| 1 | Distribution of sexually explicit images without consent |
| 2–3 | Possession or viewing of matter portraying a sexual performance by a minor (over 12, under 18) — two counts |
| 4 | Distribution of matter portraying a sexual performance by a minor (over 12, under 18) |
“We are aware of a situation at South Middle School involving a student allegedly misusing artificial intelligence (AI) and inappropriate images. Henderson Police Department is involved and is investigating the situation.”
The district says it has been in contact with the families involved and is cooperating with police. No individual official is named anywhere in this story — a choice districts make deliberately when minors are involved on both sides.
What this is not
This is not the “Cat in the Hat” AI threat trend. There is no threat here, no lockdown, no viral character, no TikTok. Nobody was trying to frighten a school. Those are different problems with different fixes, and blurring them together because both sentences contain the letters “AI” helps nobody.
What this is: a kid used a free tool on a phone to manufacture sexual images of two other kids, and then showed them around. There was no hacker. There was no stranger. The entire incident is a middle schooler, an app, and a very short afternoon.
If the charges sound heavy for a child
Sit with both halves of it. Kentucky charged a minor with possessing and distributing child sexual abuse material, because that is what the statute says those images are — regardless of who generated them or how. And two other children now have images of themselves that exist in the world. Both things are true at once, and the hard conversation this week is that your kid can become the first one without ever thinking of themselves as a criminal.
The good: Discord picked an age for your kid
Announced Tuesday, September 22 · automatic reclassification began during the week · global
Discord began automatically sorting every existing account into one of three groups. It is not asking — it is deciding, from “the communities, servers, and games you're connected to… general activity levels… account history.”
| Group | Who | What changes |
|---|---|---|
| Adult | 18+ | Nothing changes |
| Teen | 13–17 | New protections on by default; core features still work |
| Unconfirmed | Not enough signal | Some adult-content restrictions until verified |
“No single server determines your age group, and we don't look at your messages, calls, or other content to decide which age group you're in.”
A teen account gets four things by default: messages from non-friends route to a message-requests inbox; sensitive images are blurred or blocked; age-restricted servers, channels and bot commands become inaccessible; and the full profile and current activity are visible only to friends and people in smaller shared servers. Nobody has to opt in — which is what advocates have been asking platforms to do for a decade.
Discord also says “more than 90% of you won't be asked to confirm your age.” For nine out of ten families, this happened silently.
If the guess is wrong, you can fix it without handing over an ID
The options include sharing the age range your Apple App Store or Google Play account already holds, a credit-card check run through Stripe, a video selfie processed on your own device, a passport via Google Wallet in a handful of countries, a full ID scan, or a reusable credential called AgeKey. The app-store option is the easiest and involves photographing neither your face nor your documents.
Two pieces of honesty
Discord is not doing this purely out of goodwill, and says so: “age assurance laws are expanding around the world,” and it had already launched age assurance in Brazil and in Texas. And this is inference — a machine guessing your child's age from which games they play and who they spend time with. It will be wrong in both directions. A 13-year-old on an old account playing grown-up games can get read as an adult.
The bad: Spokane went dark
Spokane Public Schools, WA · incident Sunday, September 20 · families notified Monday, September 21 · still down Thursday, September 24
Spokane Public Schools — Washington's second-largest district, roughly 30,000 students across 58 schools — was notified of a network security threat on Sunday evening and told families the next morning: “Out of an abundance of caution, we have chosen to take several of our systems offline.” Down went PowerSchool (grades, attendance, parent-teacher messaging) and BusinessPlus (payroll).
Unlike Springfield, schools stayed open. Teachers took attendance by hand. Superintendent Adam Swinyard: “We did school for a very long time without technology, and so our staff are prepared to do that.” Spokesperson Ryan Lancaster on why those two systems specifically: “We took them offline because they are our most important pieces.” Four days in, the systems were still down with no restoration timeline.
The careful part
On the question every parent cares about — whose information was taken — the district's answer is “preliminary findings indicate only staff were affected by the incident.” Swinyard said it is too early to speculate on the extent of information affected or who is responsible.
Underline preliminary. Early scope findings in incidents like this get revised, and they usually get revised in one direction. This is the single most likely sentence in this week's report to change after publication.
No official has said ransomware. No official has named a group. No ransom demand has been confirmed. We did check the leak-site trackers: there is an entry logged the same day for a US education victim with the name masked, attributed to a group called Nightspire, with no data and no ransom amount. We are not going to tell you that entry is Spokane, because we can't — the name is redacted, no outlet has reported it, and the district has named no one. That is a lead, not a fact, and the difference is the entire job.
One thing worth heading off: PowerSchool was in the news in early 2025 for a very large breach of its own systems. This is not that. PowerSchool says its hosted product and other customers' data were not affected here. PowerSchool went dark in Spokane because Spokane pulled the plug on it. That is two districts in three weeks — and in both, the first public numbers cover staff.
The Decoder Ring: newly official, not new
For eight weeks this segment came back empty. On September 15 Merriam-Webster added about 1,400 new words, so the dictionary did the work instead. Be clear on what is new: these are not new words. Your kid has said them for years. The dictionary finally admitted they exist.
- Looksmaxxing — treating your own face and body as a project to be optimised. The one to pay attention to: it starts at skincare and haircuts, and there is a pipeline from there into corners where boys are told their bone structure already decided their future.
- Crashout — to lose control emotionally, to melt down. Used lightly and constantly, and occasionally not lightly at all.
- Copium — coping + opium. The comfort taken in a story that isn't true after a loss.
- Parasocial — a one-sided relationship with a creator who does not know your kid exists. The most useful word here, because a kid who has a name for something can notice it happening.
- Cuffing season — in the dictionary's own wording, “a time of the year when people who are not married or in a relationship begin looking for a short-term romantic partner to spend the colder months of the year with.”
Merriam-Webster's site blocks automated lookup, so only “cuffing season” above is verbatim. The other four are our plain-language wording.
A government sued OpenAI — and a year of phone bans got measured
On Monday, September 21, British Columbia Attorney General Niki Sharma, filing jointly with the Peace River South School District, sued OpenAI over the February 10, 2026 Tumbler Ridge school shooting that killed eight people, six of them children — naming CEO Sam Altman as a defendant. The claim: OpenAI failed to notify law enforcement about threats made in ChatGPT, and designed an unsafe product.
Sharma says the shooter's account came to OpenAI's attention in June 2025 over repeated references to gun violence, triggering an internal review that was never reported to the RCMP. She also confirms she has not read those conversations and says OpenAI refuses to disclose them — so this is a government's allegation, untested in court. OpenAI's response to reporters did not address it. It matters here because the question underneath is the question: when a chatbot notices a kid talking about hurting people, who does it owe a phone call to?
Separately, a study published Tuesday, September 22 in Frontiers in Developmental Psychology followed 960 students through a full year of a total school phone ban and found it was not associated with clear gains in social wellbeing.
It is one school, in Belgium, with no control group, so read it narrowly: taking the phones away, by itself, did not make these kids less lonely. This is not “phone bans don't work,” and the wider literature cuts both ways.
Your action plan this week
1. Check which age group Discord put your kid in.
Two minutes, together, in account settings. If your 13-, 14- or 15-year-old got read as an adult, the teen protections are off, and that is the one to fix — with the app-store age range or the on-device selfie. Don't upload a licence. And make it a conversation, not an audit: the useful question is not did you lie about your age, it is did anything change in your apps this week, and did you notice.
2. Freeze your child's credit.
Second district in three weeks — this is a pattern now. A freeze is not monitoring: monitoring tells you after someone opens a line of credit in your eight-year-old's name; a freeze stops them opening it. Free at Equifax, Experian and TransUnion, about ten minutes each, with proof you're the parent. A child's file is usually empty and nobody checks it for eighteen years — which is exactly what makes it valuable. No district can do this for you.
3. Have the AI-images conversation twice.
Once for the kid who might make one, or stand next to someone who does: it is a crime in most of this country even if you generated it, and children do get charged. Delete it, don't forward it, tell an adult. Once for the kid it might happen to: it is not your fault, you are not in trouble, and there is something we can do. That something is Take It Down, a free NCMEC service that helps pull these images from participating platforms — and it works on AI-generated images.
4. Ask what replaced the phones.
Not is there a rule. A full year of a total ban, measured at one school, produced no clear gain in wellbeing on its own; the authors' own conclusion is that restrictions need pairing with things that actively build connection. The rule is the easy half. What goes in the empty time is the half that does the work.
What we did not check this week
We check five categories every week, and we tell you when we come up short. This week the list is longer than usual, and the distinction matters: “not checked” is not the same as “checked and empty.”
- The Cat in the Hat threat trend. The story that produced more than ten juvenile arrests earlier this month got no research pass at all — not one search. We cannot tell you whether it is growing, cooling or mutating.
- Six of eight platforms. We checked Meta's newsroom (nothing on teen safety in the window) and Discord. We did not check TikTok, Snap, YouTube/Google, Roblox, OpenAI or Character.AI.
- Springfield, Massachusetts. We could not finish the local-outlet sweep. As of Friday, September 18, the breach still had not been attributed to a named group publicly, and there was still nothing about credit monitoring extending from staff to students.
- Legislation and litigation, nearly all of it: KOSA, the CHATBOT Act, the Florida AG's proposed criminal liability for chatbot makers, the FTC's companion-chatbot study (a year old this month, still no staff report), and the October bellwether in JCCP 5255.
- The New Mexico docket. The state's press releases carried nothing in the window on the $942 million child-safety judgment — no ruling, no hearing date, no bond order, no appeal. There is a jury out in Santa Fe against Meta, but it is a different case, about Cambridge Analytica and adult data privacy. We did not pull the court docket itself.
- Verbatim dictionary definitions for four of the five Decoder Ring terms. Attempted and blocked, not skipped.