What we found
The Federal Trade Commission published a consumer alert for parents describing online risks to children that include cyberbullying, scams, and digitally altered deepfakes such as "nudify" apps that create fake nude photos and videos of real people. According to that alert, the Take It Down Act requires covered platforms — including social media, messaging, and photo or video sharing apps — to provide a way to request removal of an explicit photo or video of your child, including deepfakes. The alert states that platforms must remove the image, and known identical copies, within 48 hours of the request, and points readers to TakeItDown.ftc.gov. The agency also warns that kids and teens do not always recognise the risk of sharing explicit images or deepfakes, even as a joke, and that possessing or sharing explicit images of children can be a serious crime. Agree one rule in your household now: if anyone asks for an explicit picture, video or "nudes", the answer is no and the next step is telling a trusted adult the same day. Our harvested material does not say how many children have been targeted, does not name any app or platform involved, and does not describe what happens if a platform misses the 48-hour window. The alert we reviewed describes image removal rights and online-safety advice; it does not describe any payment, extortion demand or financial loss, so we make no claim about one. LIVEFRAUD ALERTS is independent and not affiliated with the Federal Trade Commission or any platform mentioned; this alert summarises one published consumer advisory. Figures and requirements quoted here are only as current as the advisory we read.
- Single source, but it is an official Federal Trade Commission consumer alert speaking directly to the subject.
- The 48-hour removal requirement and the list of covered platform types are stated explicitly in the source text.
- No corroborating press or community reports were harvested, so scale and real-world enforcement remain uncharacterised.
- Reviewed by 3 models, 2 from independent houses.
THE RULE
Fake explicit images of a real child can be made and spread through everyday messaging and video-sharing apps; the harm is immediate, and removal depends on someone making a request.
What we don’t know
- How many children or families have been affected.
- Which apps or services have received or acted on removal requests.
- What recourse exists if a platform does not remove an image within 48 hours.
- Whether any money, extortion demand or payment method is involved in these incidents.
The bench — who voted
3 INDEPENDENT AI MODELS REVIEWED THIS. 1 OF 3 — TWO MODELS DISSENTED. READ WHY.
The card names a count. Here are the seats behind it, with what each one said.
ANTHROPICclaude-opus-5 · HIGH
The single evidence row supports only two things: that the alert describes nudify apps creating fake nude images of real people, and that the Take It Down Act requires removal of an explicit image of a child plus known identical copies within 48 hours. The finding attributes several further specifics to the alert that the row does not contain: that it lists cyberbullying and scams as risks; that 'covered platforms' are defined to include social media, messaging, and photo or video sharing apps; that it points readers to TakeItDown.ftc.gov; that kids and teens do not always recognise the risk of sharing images 'even as a joke'; and that possessing or sharing explicit images of children can be a serious crime. Each of these may well be in the real advisory, but on the harvested material as presented they are unsourced attributions to a named federal agency.
DISSENT, PRESERVEDI would record that the top-line CLAIM as worded is fully carried by the evidence; my 'overstated' verdict is directed at the expanded finding text, not at the headline. If the panel trims the unattributed specifics, this becomes a clean 'holds'.
DISSENT, PRESERVEDOn the grade I will not soften: the underlying pattern — applications generating non-consensual sexual imagery of identifiable minors — is a high-severity harm regardless of the absence of any payment or extortion element in this particular advisory. The disclosure that no financial element is claimed should not be read by anyone as lowering the stakes.
GROQopenai/gpt-oss-120b · MODERATE
The finding adds that platforms must provide a way to request removal and lists specific platform types (social media, messaging, photo/video sharing) and that the FTC warns kids may not recognize risk and that sharing explicit images is a serious crime—none of these details are present in the single evidence row provided.
OPENROUTERopenrouter/free · LOW
The claim omits that the alert does not provide enforcement data or case numbers, which could lead readers to overestimate how widespread the issue is.
Reviewed by 3 independent models: 1 found it carried by the evidence, 2 did not.
▼ Protocol & challenge record
Objection (medium): Risk_line is not supported by the harvested row and reverses the source's logic. The row lists "social media, messaging, and photo or video sharing apps" as COVERED PLATFORMS subject to removal duties — not as the vectors through which nudify images are made and spread. Writing "Fake explicit images of a real child can be made and spread through everyday messaging and video-sharing apps" converts a legal-coverage list into a threat-vector claim the FTC never made. "The harm is immediate" also inflates the source's "cause real harm."
Resolved: Rewrite risk_line to stay inside the row, e.g.: "The FTC says nudify apps can create fake nude photos and videos of real people, and that fake explicit images of children posted online cause real harm; removal depends on someone making a request to the platform." Drop the claim that messaging/video apps are how such images are made and spread, and drop "immediate."
Objection (medium): The advice sentence invents specificity absent from the source. The row's trusted-adult instruction is triggered by SEEING an explicit photo or video of yourself or someone you know; the draft re-triggers it on being ASKED for one, and adds a "same day" deadline that appears nowhere in the harvested text. Advice-role sentences still carry FTC-adjacent authority in context, and a fabricated timeframe next to a genuine statutory 48-hour figure invites readers to treat both as official.
Resolved: Replace with the source's own two instructions, kept separate: say no to anyone who asks for an explicit picture, video or "nudes"; and talk to a parent or trusted adult if you see an explicit photo or video of yourself or someone you know online. Delete "the same day."
Objection (medium): Date/provenance mismatch is unresolved. The URL path is /consumer-alerts/2026/09/ but the harvested pub timestamp is 2026-10-01T16:00:00Z. One of these is wrong, or the timestamp is a crawl/re-publication artifact rather than original publication. The draft asserts currency ("only as current as the advisory we read") without ever stating the date, so a reader cannot judge currency at all. No publication date appears anywhere in the finding.
Resolved: State the alert's date in the finding text and reconcile the URL month against the pub timestamp before publishing; if they cannot be reconciled, cite the date as it appears in the URL path and note the discrepancy in the evidence limitation.
Objection (medium): Scope narrowing risks a false negative implication. The finding describes the Take It Down Act removal right exclusively as a right to request removal of an image "of your child." The source frames it that way because it is a parenting alert, but the draft is now a standalone artifact; a reader reasonably infers the 48-hour duty applies only to minors. Nothing in the harvested row says the right is limited to children, and the draft should not let that inference stand unqualified.
Resolved: Add a scoping line to the limitations: "The alert we reviewed addresses parents and describes the removal right in relation to a child's image; it does not state whether the same 48-hour duty applies to images of adults, and we make no claim either way."
Objection (medium): The limitation sentence misstates the source. The draft says the alert "does not describe any payment, extortion demand or financial loss" — but the row explicitly contains "Did someone share, or threaten to share, an explicit photo or video of your child online?" A threat to share is the single most operationally relevant line in the source for a fraud-alert audience (it is the sextortion hook), and the draft both omits it from the finding and writes a limitation that reads as if no threat content exists.
Resolved: Add a sourced sentence: "The alert also directs parents to TakeItDown.ftc.gov if someone has shared, or threatened to share, an explicit photo or video of their child online." Then narrow the limitation to what is actually absent: no payment, money demand or financial loss is described.
Objection (medium): directive_options contains broken output: "Send this to any consider you know." This is ungrammatical, names no coherent group, and appears to be a template-fill failure. If it can reach a publish surface it is a credibility defect on its face. Relatedly, targeting_dropped cites a group label "worried" that corresponds to nothing in the row, suggesting the same upstream generator is producing junk tokens.
Resolved: Delete the malformed directive option and audit the generator that produced it and the "worried" targeting token; do not ship a directive list containing uninspected template output.
Objection (low): Internal incoherence in confidence_reasons: the third reason ("No corroborating press or community reports were harvested, so scale and real-world enforcement remain uncharacterised") is a reason AGAINST high confidence, filed under reasons for it. High is defensible for "the FTC alert says X" but not for "this is how the law operates in practice," and the record should say which proposition is held at high confidence.
Resolved: Split the confidence record: high for "the FTC alert states these requirements"; explicitly uncharacterised for scale, enforcement and real-world compliance. Move the third reason into a limiting note rather than a supporting reason.
Objection (low): watch_icons includes "person," which maps to nothing specific in the row. "video" and "text-message" at least track photo/video sharing and messaging apps named in the source; "person" is decorative and, on a child-safety alert, ambiguous.
Resolved: Drop "person" or replace with an icon that tracks a named element of the row.
Objection (low): The draft reproduces TakeItDown.ftc.gov as a destination without any independent domain verification. The string comes solely from the harvested text. There is also a separate, better-known NCMEC service branded "Take It Down" at a different domain; publishing one FTC-subdomain string with no verification step creates a small but real misdirection surface if the harvest is corrupted or the alert is stale.
Resolved: Verify the TakeItDown.ftc.gov destination resolves to an FTC-controlled host before publication, or attribute it strictly as "the alert points readers to TakeItDown.ftc.gov" without presenting it as independently checked.
Preserved dissent
ON THE RECORDI do not accept 'high' confidence as currently recorded. It is defensible only for the narrow proposition 'an FTC consumer alert states these things,' and the draft's own third confidence_reason argues against the grade it is filed under. On a single uncorroborated row with an internal date inconsistency, I would record high for the attribution and explicitly medium for the currency of the requirements described.
ON THE RECORDThe risk_line is the weakest element in the package and I think it should be cut rather than edited if time is short. Turning the statute's list of covered platforms into a sentence about how abuse images 'are made and spread through everyday messaging and video-sharing apps' is exactly the kind of source-to-claim stretch that survives editing because it sounds plausible. The FTC did not say it.
ON THE RECORDThe 'same day' instruction in the advice sentence is invented. I want it on the record that no harvested text supports a timeframe, and that placing an invented deadline in the same artifact as a real statutory 48-hour figure is a specific and avoidable hazard, not a stylistic quibble.
ON THE RECORDA directive option reading 'Send this to any consider you know.' reached an independent review stage. Whatever is decided about the substantive objections, I regard the presence of unreviewed template garbage in a child-safety artifact as a process failure worth recording separately from the content findings.
The sources
Official sourceTalking with parents and kids about deepfakes, online safety, and the Take It Down Act2026-10-01
The Federal Trade Commission alert describes nudify apps that create fake nude photos and videos of real people, and states that the Take It Down Act requires covered platforms to remove an explicit image of a child, and known identical copies, within 48 hours of a request.
Authority: official. Retrieved 2026-10-02.
Limitation: The alert is guidance and does not report case numbers, named platforms, enforcement outcomes, or any financial element.
Open the original source →Other checks
Published under standing founder pass (A9) — every claim source-mapped by the machine.
▼ What the machine checked
- ✓ Not a community submission.
- ✓ No entity is named.
- ✓ All 4 material sentence(s) map to FTC.
- ✗ groq returned "overstated"; openrouter raised 1 objection(s); anthropic returned "overstated" — published on the receipt, not blocking (A9 amendment).
- ✓ No audience band is set.
No human affirmed these. They were verified by the classifier described in Amendment A9, on 2026-10-02.
Something wrong here? Tell us and we'll correct it — corrections are published, not quietly edited.
Phishy? Send it → sharelivefraud.com/squire-it
Not affiliated with any government agency, credit bureau, bank, platform, or law-enforcement agency. Informational only — not legal or financial advice.
Naming a source is not an endorsement, and being named here is not an accusation against any company.
Powered by SquireIt™
Verify this receipt at squireit.com