What we found
The Federal Trade Commission says people reported losing $3.5 billion to imposter scams in 2025, with reported losses increasing nearly three times since 2020. Imposter scams were reported more than any other fraud category that year, making up nearly one in three fraud reports, and reached people through text, phone, email, social media, search engine results and other means. According to the FTC, some of the costliest impersonation scams start with a fake security alert, often from a bank, after which people are convinced to move money to "protect" it — with their losses often limited only by their available funds. The FTC reports nearly $1 billion lost to business impersonators, with the highest reported losses to bank impersonators, and about $920 million to government impersonators, up from $866 million and $789 million respectively in 2024. Across all fraud types, about $16 billion was reported lost in 2025, described by the agency as the highest on record and an increase of about 25% over 2024. If an alert tells you to move, transfer or "protect" your money, do not act on the message: hang up or close it, then call your bank yourself using the number printed on your own card or statement. Our harvested material does not say how many people filed those reports, which banks or agencies were impersonated in the fake alerts, or what share of the money was recovered. Two of the three rows we hold are press headlines with no article text, so we cannot check any figure they carry beyond what the FTC release itself states. The FTC says consumers targeted by a government impersonation scam should report it at ReportFraud.ftc.gov. LIVEFRAUD ALERTS is an independent consumer fraud awareness publication and is not affiliated with the FTC or with any agency, bank or company named here; brands and agencies appear only as impersonation targets already named by the official source. This alert was assembled only from the three harvested rows listed above and carries no other material.
- The central figures and the fake-bank-alert pattern come directly from a dated FTC press release in our harvested rows.
- Two independent press rows carry headlines consistent with the official figure and the rise in government imposter reports.
- Weakness: the press rows are headline-only, and the release gives no report counts, no named impersonated banks and no recovery figures.
- Reviewed by 3 models, 2 from independent houses.
THE RULE
High exposure: the FTC describes losses in these bank-alert cases as often limited only by the funds a person can reach, and reported imposter losses are at a record.
What we don’t know
- How many people filed the 2025 imposter scam reports.
- Which banks or agencies were impersonated in the fake security alerts.
- How the money moved out in these cases — transfer, cash, cards or other channels.
- How much of the reported $3.5 billion was returned to consumers.
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
The finding carries numerous specific figures that appear nowhere in the three harvested rows: 'nearly three times since 2020', '$866 million and $789 million respectively in 2024', 'about $16 billion' across all fraud types, 'highest on record' for that total, and 'an increase of about 25% over 2024'. Row [1] is summarized only as $3.5B, one-in-three reports, fake bank alerts, ~$1B business and ~$920M government impersonators. Rows [2] and [3] are headlines carrying no figures of their own (row [3] explicitly none). These numbers are therefore unverifiable against the stated evidence.
DISSENT, PRESERVEDThe core of the CLAIM — a $3.5 billion 2025 imposter-scam figure attributed to FTC reports, and the fake-bank-security-alert-then-move-your-money pattern as a costly opening move — is directly carried by row [1] and should not be struck. My objection is to the accreted figures around it, not to the central assertion.
DISSENT, PRESERVEDThe operational advice (do not act on the alert; hang up or close it; call the number on your own card or statement) is sound, actionable and independent of the disputed statistics. If the panel trims the piece, keep that line.
DISSENT, PRESERVEDI grade the pattern high regardless of the sourcing dispute: authorized-push-payment fraud initiated by a spoofed bank alert typically drains whatever the victim can reach, is often treated as customer-authorized, and recovery rates are poor. The finding itself concedes it does not know what share was recovered, which argues for a high grade rather than against it.
DISSENT, PRESERVEDRecommended remedy: cut or separately source the 2020 comparison, the 2024 comparatives, the $16 billion all-fraud total and the 25% figure, drop the bank-impersonators-are-highest ranking, and lower stated confidence to moderate.
GROQopenai/gpt-oss-120b · HIGH
The finding adds numerous quantitative claims (e.g., losses tripling since 2020, imposter scams being the top fraud category, $16 billion total fraud loss, 25% rise over 2024, specific channels of outreach, and that bank impersonators incurred the highest losses) that are not present in any of the three evidence rows provided.
OPENROUTERopenrouter/free · MODERATE
The finding states that reported losses increased nearly three times since 2020, but no evidence row supports this claim.
Reviewed by 3 independent models; all judged the finding to go beyond the evidence.
▼ Protocol & challenge record
Objection (high): "Record" is not in the official source and is attributed to the FTC. The claim line reads "FTC data show people reported losing a record $3.5 billion to imposter scams in 2025," and risk_line repeats "reported imposter losses are at a record." The FTC release never calls the $3.5B imposter figure a record — it says imposter losses "increas[ed] nearly three times since 2020," and reserves "the highest on record" for the separate all-fraud $16 billion figure. The only "record" language in the harvest is the qz.com headline (row 63e68087), a headline-only press row the draft itself says it cannot check. So the draft launders a press adjective into an attribution to federal data, and simultaneously risks bleeding the $16B "highest on record" superlative onto the $3.5B number. Note the finding body correctly omits "record" — the overclaim lives only in the claim line and risk line, which is where readers see it first.
Resolved: Strike "record" from the claim line and the risk_line, or attribute it: claim becomes "FTC data show people reported losing $3.5 billion to imposter scams in 2025 — nearly three times 2020 levels." If "record" is retained anywhere it must be tied to the $16 billion all-fraud figure, which the FTC does call the highest on record, with the press "record" wording for the imposter figure marked as a press characterisation, not FTC language.
Objection (medium): Quantifier drift from "some of the costliest" to "the costliest ... often." FTC: "Some of the costliest impersonation scams start with a fake security alert, often from a bank." Claim line: "the costliest cases often starting with a fake security alert appearing to come from a bank." The source scopes an unquantified subset; the claim converts it into a general property of the top-loss population. The finding sentence preserves the original wording, which makes the claim line the weakest sentence in the package.
Resolved: Restore the source quantifier in the claim line: "...with some of the costliest cases starting with a fake security alert appearing to come from a bank..." This matches the finding sentence, which is already correct.
Objection (medium): Date ordering breaks the corroboration story for row 1c3e279d. The WGAL row is dated 2026-05-12, five weeks BEFORE the FTC release it is being used to corroborate (2026-06-15). A May story cannot be reporting on a June release; it is either a different FTC data product, a Sentinel dashboard read, or an RSS date artifact. Either way, confidence_reasons #2 ("Two independent press rows carry headlines consistent with the official figure and the rise in government imposter reports") cannot rest on it as downstream confirmation of this release.
Resolved: Note the pre-dating in the evidence limitation for 1c3e279d ("published 2026-05-12, before the 2026-06-15 release; cannot be downstream reporting on it") and remove it from confidence_reasons as corroboration of the June figures.
Objection (medium): Internal contradiction on what row 1c3e279d supports. The evidence block states it "gives no figures of its own," and the limitation says headline-only rows cannot be checked for any figure — yet finding_sentences cites 1c3e279d as a row_id for the sentence carrying four specific dollar amounts ($1B business, $920M government, $866M/$789M for 2024). A headline reading "significant rise in 2025" does not support any of those numbers. The citation should be removed or the sentence split so the headline only backs the directional "rise" language.
Resolved: Drop 1c3e279d from the row_ids of the four-figure sentence; leave that sentence sourced to the FTC release alone. If the headline is used at all, use it only alongside a directional statement about the government-imposter rise.
Objection (medium): The limitation "our harvested material does not say ... what share of the money was recovered" is inaccurate as written. The harvested FTC release does carry a recovery figure: "a dozen enforcement actions ... obtained over $70 million in redress for consumers" under the Impersonation Rule since 2024. That is not a share of the 2025 $3.5B and should not be presented as one, but the draft currently tells readers the harvest is silent on recovery when it is not. The honest framing is: the harvest reports $70M+ in redress across Impersonation Rule cases since 2024, which cannot be mapped onto the 2025 imposter total — a materially different and more useful statement, and one that quietly signals how little of $3.5B comes back.
Resolved: Rewrite the limitation to say the harvest reports over $70 million in redress obtained across a dozen Impersonation Rule enforcement actions since 2024, and that this figure cannot be mapped to the 2025 $3.5 billion total, so the share of 2025 losses recovered remains unknown. Move "share of 2025 losses recovered" to unknowns and remove the false claim of total silence.
Objection (medium): "Two independent press rows" overstates independence. Both press rows are Google News RSS redirect URLs (news.google.com/rss/articles/CBMi...), not resolvable publisher URLs, with no article text harvested. Neither can be shown to have done independent verification; the qz.com headline is almost certainly a straight write-through of the same FTC release. Two derivative headlines off one primary source are one source, not three. This inflates the confidence rationale even though the underlying primary source is strong on its own.
Resolved: Change confidence_reasons #2 to state that the two press rows are headline-only Google News redirects that appear to derive from the same FTC release and add no independent verification; rest the confidence on the primary official row.
Objection (low): Unsourced advice where a sourced alternative existed. The advice sentence (hang up, call the number on your card/statement) carries no row_ids. It is sound generic practice, but the harvested release points to a specific FTC-endorsed resource — the EJCC "Never Ever" campaign and ejcc.acl.gov/imposters, plus the FTC's framing of actions government and business will "never ever" take. Using unsourced advice while leaving an in-evidence official resource on the table is a missed opportunity and leaves the only action step in the piece resting on nobody's authority.
Not resolved — preserved on the record.
Objection (low): Confidence "high" is defensible for the core figures but is being applied to the whole package, including the "record" descriptor (O1), the quantifier shift (O2), and a corroboration claim that does not hold (O3, O4, O6). Confidence should be scoped: high on the FTC-stated figures and the fake-bank-alert pattern; not high on "record," and no lift at all from the two headline rows.
Resolved: Keep confidence "high" only if O1, O2, O4 and O6 are all applied; otherwise downgrade to medium-high with the scope stated explicitly (high on FTC-stated figures, no lift from press rows, "record" unsupported).
Objection (low): Alternative explanation not flagged for the year-over-year comparisons. The $866M→$1B and $789M→$920M increases are increases in *reports to the FTC*, which move with reporting propensity, campaign publicity (the FTC ran the Never Ever campaign and EJCC outreach) and referral-partner feeds, not necessarily with underlying victimization. The draft's evidence limitation notes "reported losses, not confirmed totals" but the finding never tells the reader that a rise in reports is not by itself a measured rise in fraud.
Not resolved — preserved on the record.
Preserved dissent
ON THE RECORDI do not accept "record" in the claim line. The FTC did not describe the $3.5 billion imposter figure as a record; the word comes from a headline-only press row that the draft itself concedes it cannot check. Attributing that adjective to "FTC data" is the single clearest source-to-claim stretch in this package, and it sits in the most-read sentence. If the Desk keeps it, the record should show I objected.
ON THE RECORD"Two independent press rows" is not a sustainable description of two unresolvable Google News redirect headlines with no article text, one of which predates the release it supposedly corroborates. They are decoration, not evidence, and the confidence rationale should not cite them as support.
ON THE RECORDThe limitation claiming the harvest says nothing about recovery is factually wrong against our own harvested text, which reports over $70 million in redress. Publishing a limitation that our primary source contradicts is worse than publishing no limitation at all, because it invites a correction on the one part of the piece meant to demonstrate care.
ON THE RECORDI would not sign this at confidence "high" as currently written. The underlying primary source is strong; the packaging around it is not, and "high" should not be used to cover the claim-line adjectives and the phantom corroboration.
The sources
Official sourceFTC Data Show People Reported Losing $3.5 Billion to Imposter Scams in 20252026-06-15
FTC press release states $3.5 billion in reported imposter scam losses for 2025, nearly one in three fraud reports, fake bank security alerts as a costly opening move, ~$1 billion to business impersonators and about $920 million to government impersonators.
Authority: official. Retrieved 2026-08-18.
Limitation: Reports losses as reported to the FTC, not confirmed totals; gives no number of reporting consumers and names no impersonated bank.
Open the original source →ReportedGovernment imposter scams | FTC reported a significant rise in 2025 (WGAL)2026-05-12
Press headline reports the FTC recorded a significant rise in government imposter scams in 2025.
Authority: press. Retrieved 2026-08-18.
Limitation: Headline only; gives no figures of its own and no detail on the rise.
Open the original source →ReportedAmericans lost a record $3.5 billion to imposter scams in 2025, FTC data show (qz.com)2026-06-26
Press headline reports Americans lost a record $3.5 billion to imposter scams in 2025 on FTC data.
Authority: press. Retrieved 2026-08-18.
Limitation: Headline only in our harvest; no article body to check how the figure was handled.
Open the original source →Other checks
Approved by ihubglobalhq on 2026-08-18, after the six-point evidence checklist.
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