What we found
The Federal Trade Commission announced on 24 September 2026 that it is considering whether to update its Rule on Impersonation of Government and Businesses, or take other action, to stop online platforms from ad-optimization practices that may be furthering impersonation scams. In 2025 the FTC received more than 1 million reports about imposter scams, with consumers reporting losses of nearly $3.5 billion. The FTC also states that nearly 30% of consumers who reported losing money to scammers in 2025 said they were first contacted on a social media platform, with reported losses of $2.1 billion. The Bureau of Consumer Protection Director is quoted saying today's impersonation scams are no longer isolated con artists but operations powered by the same advertising and targeting tools platforms sell to legitimate businesses. The FTC says it will never demand money, make threats, tell you to transfer money, or promise you a prize. Never reach a company or a government agency through an ad, a sponsored search result or a social media post — close it, type the official address yourself, and use the contact details on that site instead. The FTC notice does not name any individual platform, advertiser or impersonated business, and it does not describe any single scam script or message. The harvested material gives national totals only and does not break the reports down by age, income or location, so we cannot say who was targeted most. LIVEFRAUD ALERTS is independent and not affiliated with the FTC or with any platform referred to in this alert; this alert was written only from the published notice cited above.
- Single official FTC press release, quoted figures and dates taken directly from it.
- Loss totals, report counts and the 30% social-media contact figure are stated explicitly by the agency.
- No independent corroboration was harvested, and the rulemaking is only at the advance-notice stage — no rule has changed.
- Reviewed by 3 models, 2 from independent houses.
THE RULE
An ad or sponsored result inside a platform you already trust can be a scammer wearing a real company's or agency's name.
What we don’t know
- Which platforms, ad tools or advertisers the FTC is concerned about — none are named.
- What the impersonation ads actually look like, or what they ask consumers to do.
- How the money moved in these reported losses, and whether any of it was recovered.
- Whether the Commission will amend the existing rule, propose a new one, or take non-regulatory measures.
The bench — who voted
3 INDEPENDENT AI MODELS REVIEWED THIS. 2 OF 3 — ONE MODEL DISSENTED. READ WHY.
The card names a count. Here are the seats behind it, with what each one said.
ANTHROPICclaude-opus-5 · HIGH
Only one evidence row was supplied, and it covers only the fact that the FTC is considering updating its impersonation rule. Every quantitative claim in the finding — more than 1 million imposter-scam reports, nearly $3.5 billion in losses, nearly 30% first contacted on social media, $2.1 billion in social-media losses — has no evidence row behind it. The CLAIM headline leads with the $3.5 billion figure, so the most attention-grabbing number in the alert is entirely unsourced in the material provided.
DISSENT, PRESERVEDThe core rulemaking statement is accurately reported and properly hedged, and the disclosure paragraph about unnamed platforms and national-totals-only is unusually honest. My objection is to the unsourced statistics and the quote, not to the alert's premise.
DISSENT, PRESERVEDI grade the underlying pattern high regardless of this alert's sourcing problems. Sponsored search results and social ads that impersonate a bank, a retailer or a government agency reach consumers at the exact moment they are looking for a legitimate contact number, and the platform's own placement lends them borrowed credibility. The single behavioural instruction — never use the ad, navigate to the official site yourself — is the right one and would survive even if every number in this alert were wrong.
DISSENT, PRESERVEDIf this is published, the numbers should either be given their own sourced rows or cut entirely. Publishing a billion-dollar figure with no row behind it invites exactly the kind of challenge that discredits otherwise correct warnings.
GROQopenai/gpt-oss-120b · HIGH
The finding adds numerous factual claims—2025 loss totals of $3.5 billion, 1 million reports, 30% first contact via social media, $2.1 billion loss figure, quoted Bureau Director remarks, and detailed consumer advice—that are not present in the single evidence row, which only states the FTC is considering rule updates.
OPENROUTERopenrouter/free · MODERATE
The sole evidence row supports only that the FTC announced consideration of a rule update or other action concerning ad-optimization practices that may further impersonation scams. It does not substantiate the reported-loss totals, social-media percentages, the quoted characterization by the Bureau Director, or the FTC safety statements included in the finding.
DISSENT, PRESERVEDThe core first sentence is directly carried by the evidence row; the overstatement lies in the extensive additional statistics, quotation, advice, and methodological claims appended to it.
Reviewed by 3 independent models; all judged the finding to go beyond the evidence.
▼ Protocol & challenge record
Objection (high): Person-name leak in the targeting machinery. The source's only 'Christopher' is Christopher Mufarrige, Director of the Bureau of Consumer Protection. The generator has turned that into a demographic: callout option 'ATTENTION: CHRISTOPHERS' and two directive options ('Send this to any christopher you know', 'Forward this to the christophers in your life'). That is a named individual from the byline of the release being converted into a supposed victim class. Nothing in the row supports people named Christopher being targeted. Had either directive been selected, the alert would have manufactured a targeted group out of an FTC official's first name. The same failure produced 'ATTENTION: VETERANS' (no veteran, military or benefits content anywhere in the row) and 'ATTENTION: SCAMMEDS' (not a word). The final selections happen to be clean, but the option set shows the targeting extractor is pulling arbitrary tokens, and 'targeting_dropped' confirms it — it dropped 'through' and 'links' as if they were groups of people. This needs to be recorded even though the shipped output avoided it.
Resolved: Record in full on the public log. The shipped callout and null share_directive contain no name leak, so nothing in the published alert changes; the objection stands against the option generator, not the output.
Objection (medium): Double-counting hazard between the two money figures. The finding places '$3.5 billion to imposter scams' and '$2.1 billion' from social-media-first-contact scams in adjacent sentences with no note that they measure different populations. The $2.1bn is losses across ALL scam types where first contact was on social media — not impersonation losses, and it overlaps the $3.5bn to an unknown degree. A reader scanning the alert can easily read them as additive ($5.6bn) or as two cuts of the same imposter total. The release does not permit either reading.
Resolved: Add a qualifier to the third finding sentence: '...said they were first contacted on a social media platform, across all scam types, with reported losses reaching $2.1 billion — a figure that overlaps the imposter total and cannot be added to it.'
Objection (medium): The 'Never' advice line is unsourced, absolute, and sits inside a block otherwise attributed to the FTC. 'Never reach a company or a government agency through an ad, a sponsored search result or a social media post' carries no row_ids — the FTC release gives no such guidance. It is wedged between two sentences that are explicitly FTC statements ('The FTC also states...', 'The FTC says it will never demand money...'), so an ordinary reader will attribute the imperative to the agency. Either attribute it to LIVEFRAUD explicitly or move it out of the FTC-attributed run of sentences.
Not resolved — preserved on the record.
Objection (medium): The 'bank' watch icon is not in evidence and contradicts the draft's own unknowns. The unknowns list says 'How the money moved in these reported losses' is unknown; the row describes no payment method, no bank transfer, no wire, no account takeover. Flagging a bank icon tells the reader a payment channel that the draft has just conceded it cannot identify. Drop it or justify it from the row.
Resolved: Remove 'bank' from watch_icons; retain 'link' and 'person', both of which map to the ad/impersonation content actually in the row.
Objection (medium): The audience callout narrows the exposure wrongly. 'ATTENTION: ANYONE CONTACTED ON SOCIAL MEDIA' is drawn from the 30% first-contact statistic, but the ANPRM's subject is ad-optimization tools at 'social media, search engines and other digital marketplace platforms.' The largest single channel in the rulemaking — sponsored search results — is excluded by the callout, even though the draft's own advice line warns about sponsored search results. A reader who only uses search and marketplaces is told by the callout that this is not about them.
Not resolved — preserved on the record.
Objection (medium): Omitted material fact: the vote was 2-0, and the comment window is 60 days after Federal Register publication. The 2-0 vote on a five-seat Commission is context a reader is entitled to when judging how settled this direction is. The comment deadline is the single actionable thing a consumer can do with this notice, and it has been dropped entirely while an unsourced 'never click' instruction was added in its place.
Not resolved — preserved on the record.
Objection (low): Quote handling. The draft paraphrases the Bureau Director's quote without naming him, while the source names Christopher Mufarrige. Anonymising a named, on-the-record official weakens verifiability for no gain, and the paraphrase also drops 'sophisticated, highly engineered,' which is the load-bearing adjective in the quote. Separately, the draft omits his caveat 'and the true cost is likely far higher,' which is the agency's own signal that $3.5bn is a floor.
Resolved: Name the speaker: 'Christopher Mufarrige, Director of the FTC's Bureau of Consumer Protection, is quoted saying...' and restore the 'true cost is likely far higher' caveat alongside the $3.5bn figure.
Objection (low): Confidence 'high' is defensible for a primary official source but is internally awkward: the third confidence_reason states there is no corroboration and no rule has changed, which is a reason for caution, not for high confidence. It reads as a limitation filed under the wrong heading. The rating is fine; the reasoning should not list a counter-indicator as a support.
Not resolved — preserved on the record.
Objection (low): Minor source-to-claim drift on wording: the release says social-media losses 'reaching $2.1 billion'; the draft says 'of $2.1 billion.' 'Reaching' signals an upper bound of a range in FTC's usage; 'of' states a point figure. Small, but it is the kind of tightening that should go the other way.
Resolved: Change 'with reported losses of $2.1 billion' to 'with reported losses reaching $2.1 billion', matching the source verbatim.
Objection (low): The headline claim's 'after consumers reported nearly $3.5 billion lost to imposter scams in 2025' implies the ANPRM was issued in response to that figure. The release cites the figure as background; it does not state the data triggered the rulemaking. A post-hoc framing the source does not assert.
Resolved: Recast the claim as 'The FTC is asking whether online platforms' ad-optimization tools help scammers impersonate real businesses and government agencies. The agency cites nearly $3.5 billion in consumer-reported imposter-scam losses in 2025.' Two statements, no implied trigger.
Objection (low): Unexamined alternative explanation. The 30% social-media first-contact statistic is consistent with social media simply being where people spend time, not with ad-optimization tools causing the losses. The FTC itself only asks the question — the ANPRM 'seeks comment on the extent to which' these tools further scams. The risk_line ('An ad or sponsored result inside a platform you already trust can be a scammer wearing a real company's or agency's name') asserts the mechanism as established fact. It is plausible and broadly true, but it is not what this row establishes, and the alert's only evidence row is an open question.
Not resolved — preserved on the record.
Preserved dissent
ON THE RECORDThe single most serious thing on this record is not in the published text at all: the option generator offered 'ATTENTION: CHRISTOPHERS' and 'Send this to any christopher you know' after ingesting a release whose only Christopher is the FTC official who signed the quote. Combined with 'ATTENTION: VETERANS' (zero basis in the row) and with 'through' and 'links' being processed as candidate groups of people, this is not a near-miss on one alert — it is evidence that the targeting extractor does not distinguish a victim class from an arbitrary token in the text. The clean final selection here is luck, not control. I want this preserved verbatim.
ON THE RECORDI do not accept 'bank' in watch_icons while the draft's own unknowns state that how the money moved is unknown. An alert cannot flag a payment channel in the same document where it concedes it cannot identify one.
ON THE RECORDThe finding's only evidence row is an advance notice asking a question. The risk_line answers that question in the affirmative on the FTC's behalf. I think that is an overclaim, and the fact that it is a widely believed overclaim does not make it sourced.
The sources
Official sourceFTC Seeks Public Comment on Whether to Update Rule on Impersonation of Government and Businesses to Address Platforms’ Role in Promoting Impersonation Scams2026-09-24
The Federal Trade Commission announced on 24 September 2026 that it is considering whether to update its Rule on Impersonation of Government and Businesses, or take other action, to stop online platforms from ad-optimization practices that may be furthering impersonation scams.
Authority: official. Retrieved 2026-09-25.
Limitation: An advance notice of proposed rulemaking seeking public comment, not a finding against any named platform; the release names no platforms, advertisers or impersonated brands and describes no specific scam message.
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.
- ✗ Draws on an FTC enforcement release, which names a defendant: "FTC Seeks Public Comment on Whether to Update Rule on Impersonation of Governmen".
- ✓ All 5 material sentence(s) map to FTC.
- ✗ anthropic returned "overstated"; groq returned "overstated"; openrouter 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 2026-09-25.
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