Parent Pulse: Weekly Radar
August 17 – August 22, 2026 | Digital Youth Culture & Safety Report
Fourth week running with no new challenge and no new slang. That is the finding, not a gap. What happened instead was a courtroom — opening statements against Meta in Oakland, in front of an advisory jury — and a statistic that has frightened parents all year turning out to be a checkbox on a form.
This report discusses sextortion and child sexual abuse imagery statistics. No method or circumstance is described, and every untested claim is presented as an allegation. 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
105 extracted claims went through three-vote adversarial verification; 25 reached a verdict, 21 were confirmed and 4 were killed. Of everything confirmed, only three items are events inside August 15–21. The rest is background or correction — and the corrections are, again, the most useful thing here.
All five categories were researched this week. Last week our influencer and creator angle failed mid-run and we said so on the record. It ran this time, and its result is a genuine — if bounded — zero. Every zero on this page is a checked zero.
One claim we carried last week is now confirmed. We reported that the Oakland jury is advisory before any source we could reach said so outright. Courtroom reporting from August 18 now states it plainly, and we have cited it below.
🏃♂️ 1. Viral Challenges & Trends
Nothing cleared verification — fourth week running
No dare, stunt, or challenge-format trend originated in the August 15–21 window. Every candidate traced back to an outlet re-reporting something older.
What the dateline check disqualified. The most-shared “door-kick challenge” article carries the dateline August 17, 2025 — exactly one year outside the window — and it self-describes the trend as “again making the rounds.” The nearest 2026 item, a county prosecutor's warning, is dated August 14, one day before the window opens; earlier recurrences are from June 22. A separate medicine-misuse story dates to June 19, 2026.
We describe the mechanics of neither. Both are cases where the method is the danger, and no official investigating body has attributed a death in any of this coverage to an online challenge.
🗣️ 2. Trending Slang — and What We Decoded Instead
Fourth week running with no new slang. We looked. There isn't any. Rather than invent a list, we decoded what this week's news is getting wrong — because a parent acting on a wrong definition is worse off than a parent who never heard the word.
“764”
Not a code word. A network. It keeps appearing on lists of “codes parents should know.” It is neither a code nor a word — it is the name of a violent online criminal network. FBI Boston's February 19, 2026 advisory describes more than 350 subjects under investigation nationwide, with all 56 field offices involved. Scanning a child's texts for the digits means looking for the wrong thing entirely.
“Six-Seven”
Not the 2026 Word of the Year. Dictionary.com named “67” its Word of the Year for 2025, on October 28, 2025. No 2026 designation exists yet — those run October to December. The viral article about teachers dreading the “2026–27 school year” was published in December 2025; the 2026–27 is the school year, not the date.
“Six-Seven” as a coded reference
We have now seen it described as a drug code, a sexual act, and a self-harm reference. Dictionary.com's own entry calls it “impossible to define” and “meaningless.” It appears on no DEA slang list. It came from a rap track and a hand gesture.
“Advisory Jury”
A jury whose findings guide the judge, who may accept or set them aside. Eight people are seated in Oakland. This is why an October headline reading “jury finds against Meta” will be describing a recommendation, not a ruling.
“Check the dateline”
The five-minute media-literacy skill that disqualified three stories this week. Find the date on the alarming article, then ask: is this from this week, or from last year going around again? One item failed by exactly 365 days.
📸 3. Influencers, Creators & Kidfluencers
Researched this week — and the result is a checked zero
This category's research pass failed entirely last week and we disclosed that rather than implying it had been checked. It ran this time. Nothing in this category broke between August 15 and 21.
California SB 1247, “Social media platforms: child influencers” (Sen. Steve Padilla, SD-18), is real and live, and it is the thing most likely to be misreported. We queried the Legislature's own bill-history database directly on August 21. Its most recent recorded action is August 10, 2026 — five days before this window opens — “Read third time and amended” and “Ordered to third reading” in the Assembly. It passed the Senate 39–0 on May 19.
Status: PENDING. Not chaptered. Not enacted. If anyone tells you California just passed a law protecting kid influencers, they are ahead of the facts. California's house-passage deadline falls at the end of August, so this could genuinely move in the week of August 24 — we will re-check before the next episode.
Bounded null. Only California was examined. Kidfluencer statutes in Illinois, Minnesota, Utah and Washington, platform payout-policy changes for minors, and influencer-misconduct prosecutions were not researched.
⚖️ 4. Apps, Regulation & Litigation
Opening statements — Oakland. On Tuesday, August 18, 2026, opening statements were delivered in the U.S. District Court for the Northern District of California before Judge Yvonne Gonzalez Rogers. The states' counsel summarized Meta's business model in four words: hook, hold, harvest, hide, alleging Meta designed Facebook and Instagram to keep children on the platforms longer and longer “to the point of physical and mental harm.”
Get the count right — most headlines don't
29 states sued Meta as a coalition in 2023, but only four are trying this case: California, Colorado, Kentucky and New Jersey. Not 29 lawyers at the table — four states, going first. One national outlet's headline this week read “29 AGs testify against Meta,” which is wrong twice: attorneys general are not the witnesses, and only four states are litigating.
The detail almost nobody is reporting — now confirmed
Gonzalez Rogers convened an eight-person advisory jury, and the verdict and the penalties ultimately rest with her. So when a headline lands in October reading “Jury finds Meta liable” — or “Jury clears Meta” — that is a recommendation, not a binding verdict.
The first witness was Arturo Béjar, a former Facebook engineering director turned whistleblower; over the preceding weekend the judge rejected Meta's attempt to keep him off the stand. The trial runs six to eight weeks, and both Mark Zuckerberg and Instagram head Adam Mosseri are expected to testify.
Everything the states allege remains unproven and will be contested; an opening statement is not evidence. Sources: California DOJ press release (Aug 17, 2026); The Oaklandside, NPR, AP and CNBC (Aug 18, 2026).
New Mexico — $942 million, and a motion to freeze it
The number is $942 million, not $567 million. It breaks into a $375M civil penalty from the Phase 1 jury (March 24, 2026 — 75,000 willful violations of the NM Unfair Practices Act at the $5,000 statutory maximum, each) plus a $567M Phase 2 abatement award. Final judgment was entered August 6, 2026 by Judge Bryan Biedscheid. The widely repeated flat “$567 million” is only the abatement component.
The order also imposes five years of injunctive changes: no overnight push notifications to under-18 users (10 p.m.–7 a.m.), mandatory time-use limits for minors, public ‘like’ counts hidden by default, blocking minors from sending or receiving policy-violating nude images, a Meta-funded court-appointed Child Safety Monitor, and semiannual public compliance reports filed with the court.
None of it is in effect
On August 18 — the same day opening statements began in Oakland — Meta moved in Santa Fe District Court to stay enforcement of the entire judgment pending appeal, offering to post a $1.8 billion bond. That motion is pending; the court has not ruled, and as of Thursday, August 20, Meta had not yet filed its actual appeal (it has 30 days from August 6). Do not change how you supervise a child's phone on the strength of an order that is currently frozen.
Also in-window, and easy to over-read. On August 17, New Mexico Attorney General Raúl Torrez was reported to be working with state lawmakers on two draft bills — consumer protection and online child safety, extending to AI and chatbots — one of which would remove the cap on consumer-protection penalties. These are drafts that have not been introduced, expected to be announced “in the coming weeks.” Reported, not filed; not law.
🚩 5. Scams, Sextortion & Online Safety
No federal advisory landed inside this window
The nearest FBI IC3 public service announcement (I-081026-PSA, “Sexual Exploitation Actors Stealing and Leaking Explicit Content”) is dated Monday, August 10, 2026, five days before the window opens, and nothing newer supersedes it. Local coverage during the week was aggregator recycling of that advisory, not a new development.
The standing guidance is still the useful part. NCMEC's 2025 data release (published June 25, 2026) counts more than 50,000 reports of financially motivated sextortion in 2025 — roughly 137 per day, up from more than 36,000 in 2024 — with 94% of public reports coming from child victims or parents.
Say this precisely, because the precision matters
NCMEC advises that a victim should get help before deciding whether to pay or comply — because, in their words, “cooperating or paying rarely stops blackmail.” That is not a flat “never pay” rule, and we are not going to upgrade it into one. The point is that the decision should never be made alone, at speed, by a frightened teenager.
Note on the 50,000 figure: it counts reports received, not confirmed victims.
🔍 6. Corrections — Widely Repeated, and Wrong
1. “400,000 AI-generated child abuse reports.” This is the one worth changing your mind about. In a letter dated January 29, 2026, Stanford researcher Riana Pfefferkorn examined CyberTipline reports with the “Generative AI” box checked — roughly 485,000 in the first half of 2025 — and found that at least 78% of them involved no AI-generated material at all. The bulk came from a single company scanning AI training data and hash-matching known abuse imagery of real children, which it then correctly reported. The AI connection was that the images were sitting inside the training data — not that AI produced them.
That checkbox is a flag on a form, not a measurement. Separately, NCMEC's own 2025 annual figure of “more than 400,000 reports with a GAI nexus” includes only more than 182,000 involving offenders possessing, generating or attempting to generate that material.
None of this means AI-generated abuse imagery is not real. 182,000 is horrifying on its own — and when advocacy numbers inflate, parents quietly start discounting all of them.
2. “764 is a code.” It is a violent coercion network documented by the FBI. See section 2.
3. “Six-seven is the 2026 Word of the Year.” It is 2025's, announced October 28, 2025.
4. “Texas sued Roblox and Discord this week.” A search engine told us so. We pulled both press releases directly: the Roblox suit is dated November 7, 2025 and the Discord suit May 22, 2026. The August dates were links in a sidebar of unrelated stories. Both suits are real; neither is from this week.
5. A weekday that didn't exist. An automated summary dated the New Mexico stay motion to “Aug 19” from an article that said “Tuesday.” August 21, 2026 is a Friday, so Tuesday was August 18; the 19th was a Wednesday. We publish that one because it is our own near-miss, and it is the same class of error as the other four.
✅ What to actually do this week
1. Set the overnight rule yourself, tonight. A court decided that silencing notifications to under-18s between 10 p.m. and 7 a.m. mattered enough to order it. That order is frozen pending appeal. You are not frozen. Every phone made in the last decade can do this, and it takes about 90 seconds.
2. Retire the 400,000 number, and say why. The figure to use is 182,000, and it is still awful. Correcting an inflated statistic is not downplaying the problem — it is what keeps people believing the real numbers.
3. Teach the dateline check. Pull up any alarming story about kids and tech, find the date, and ask whether it is from this week or from last year going around again. We disqualified three stories on that test this week alone, including one that was exactly a year old to the day. Your child will be better at this than you within a month.
4. Have the sextortion conversation before you need it. NCMEC's guidance is precise: get help before deciding whether to pay or comply, because cooperating rarely stops blackmail. Say the whole sentence out loud: “If someone ever has a picture of you and is threatening you with it, you come to me first, and you will not be in trouble.” The threat only works while they think they are alone with it.
Take It Down (free, NCMEC): takeitdown.ncmec.org · NCMEC CyberTipline: 1-800-843-5678 · 988 Suicide & Crisis Lifeline — call or text 988 (US), 24/7, free and confidential.
🔎 What we did not check this week
Stated plainly, so “we found nothing” is never confused with “we didn't look”:
- Whether any state law took effect during August 15–21, beyond California SB 1247.
- Kidfluencer statutes in Illinois, Minnesota, Utah and Washington, and platform payout-policy changes for minors.
- Any FTC enforcement action inside the window.
- Platform product announcements — a targeted search found nothing dated August 15–21, but that is a bounded check, not an exhaustive newsroom sweep of Apple, Google, Meta, TikTok, Roblox, Discord and Snap.
- The 29-vs-33 state count was reconciled against reporting, not against the operative complaint.