What we found
The IRS issued a warning (IR-2023-196, Oct. 23, 2023) that criminals are soliciting donations while falsely posing as legitimate charities during global crises and natural disasters. According to the advisory, fake charity promoters may use emails, fake websites, or alter or "spoof" their caller ID so it looks like a real charity is calling. The IRS states these operations seek both money and personal information, which can be used to further exploit victims through identity theft. The advisory says criminals often target seniors and groups with limited English proficiency, and that scammers frequently use names that sound like well-known charities. The IRS says any charity asking for donations by gift card numbers or by wiring money is a scam, and that paying by credit card or check is safest — and only after confirming the charity is real. The IRS also notes donations to a fake charity cannot be deducted on a tax return, because only gifts to a qualified tax-exempt organization recognized by the IRS count. Before you give a cent, ask for the charity's exact name, website and mailing address, then look that name up yourself in the IRS Tax-Exempt Organization Search at IRS.gov. If the caller will not wait for you to run that check, hang up and give later on your own terms — a real charity will still accept your money tomorrow. We do not know how many donors have lost money to these approaches, how much was taken, or which specific fake charity names are circulating — the advisory names none. Our harvested material contains a single official IRS release and no case reports, investigator statements or complaint figures to sit alongside it. LIVEFRAUD ALERTS is an independent consumer fraud awareness publication and is not affiliated with the IRS, the FBI or any charity named or described here.
- The single source is a dated, numbered official IRS news release (IR-2023-196) quoting the IRS Commissioner by title.
- Every claim sentence restates method and guidance stated plainly in that release, with no inference beyond its text.
- Confidence is limited to the advisory's own content: it supplies no loss totals, complaint counts or named fake charities, so no scale claim is made.
- Reviewed by 2 models, 1 from independent houses.
- This is a higher-risk finding. Our rule asks for 2 independent reviews before we state full confidence; 1 returned.
- Confidence is shown as moderate for that reason, not because the evidence is weaker.
THE RULE
A donation made under pressure to an unchecked charity can cost you the money, your personal details and the tax deduction all at once.
What we don’t know
- How many people have been approached or have lost money to these fake charity solicitations.
- The dollar amounts involved.
- The specific names, websites or phone numbers being used by the fake charities.
- Whether any particular crisis or disaster appeal is being impersonated most often.
- Whether any arrests or enforcement actions have followed.
The bench — who voted
2 INDEPENDENT AI MODELS REVIEWED THIS. 1 OF 2 — ONE MODEL DISSENTED. READ WHY.
The card names a count. Here are the seats behind it, with what each one said.
ANTHROPICclaude-opus-5 · MODERATE
Several statements are attributed to the advisory that do not appear in the single evidence row supplied. The row covers impersonation, email/fake website/spoofed caller ID, gift card and wire requests, and targeting of seniors and limited-English-proficiency groups. It does not support the finding's further attributions: that the operations seek personal information for identity theft, that scammers use names resembling well-known charities, that credit card or check is 'safest', that donations to a fake charity are not deductible, or the direction to the IRS Tax-Exempt Organization Search. These may well be in IR-2023-196, but on the record before the panel they are unverified paraphrase presented as quotation of the source.
DISSENT, PRESERVEDI would grade this pattern high rather than moderate for the specific populations the advisory names. Spoofed caller ID defeats the ordinary reader's main authentication cue, gift card and wire rails make loss irreversible, and any personal data handed over compounds into identity theft long after the donation is gone. For an isolated senior or a donor operating in a second language under disaster-appeal urgency, the defence the IRS recommends is not realistically available in the moment. I grade moderate only because the countermeasure — hang up, look the name up yourself, give tomorrow — is simple, free and correctly stated in the draft.
DISSENT, PRESERVEDThe absence of any named fake charity is a genuine limit on the alert's practical utility, not a mere caveat. A reader cannot pattern-match against a list that does not exist, so the piece must live or die on the behavioural rule it teaches. The draft's instruction to hang up on anyone who will not wait for a verification check is the single most valuable line in it and should not be edited down.
GROQopenai/gpt-oss-120b · HIGH
Finding adds that criminals target "global crises and natural disasters" – the evidence row does not mention this context.
Reviewed by 2 independent models; all judged the finding to go beyond the evidence.
▼ Protocol & challenge record
Objection (high): Date currency vs. present-tense framing. The sole source is IR-2023-196, dated Oct. 23, 2023 — well over two years old — and it is explicitly pegged to the crises then 'happening around the globe.' The claim line is written in the live present tense ('Criminals ... are soliciting', 'the IRS warns') with no date anchor, so a reader of a publication branded 'LIVEFRAUD ALERTS' will reasonably read this as a current, active campaign. Nothing in the harvest establishes that the advisory is still the IRS's operative guidance, that it was not superseded, or that any of this is happening now. The claim should carry the date on its face ('In an Oct. 2023 advisory, the IRS warned...') or be re-verified against a current IRS page.
Not resolved — preserved on the record.
Objection (medium): Deduction claim drops a material condition. The source says taxpayers 'can claim a deduction if they itemize deductions.' The finding renders this as 'donations to a fake charity cannot be deducted on a tax return, because only gifts to a qualified tax-exempt organization recognized by the IRS count' — silently removing the itemization gate. This makes the tax loss sound universal when for the large majority of filers who take the standard deduction there is no deduction to lose.
Resolved: Insert the source's condition: '...cannot be deducted, and deductions are available only to filers who itemize and only for gifts to an IRS-recognised qualified tax-exempt organization.'
Objection (medium): Risk line overclaims a personal loss. 'can cost you the money, your personal details and the tax deduction all at once' asserts a deduction the reader may never have had (see OBJ-2), and asserts loss of personal details as a certainty rather than a possibility ('Typically, they seek...'). The stacking of three losses is rhetorical amplification beyond what one generic advisory supports.
Resolved: Rewrite risk line without the assumed deduction and without certainty of data loss, e.g. 'A donation made under pressure to an unchecked charity can cost you the money and the personal details you hand over — and it buys you no tax deduction.'
Objection (medium): Confidence label conflates source authenticity with claim verification. The confidence_reasons argue the release is genuine, numbered and quoted — that supports 'the IRS said this,' not 'this is happening.' The advisory itself cites no data: no complaint counts, no losses, no named entities, no case. 'High' confidence on a behavioural claim resting on one agency's unquantified characterisation, with an explicit acknowledgement that no corroborating material exists, is too strong. Medium is the honest label for the phenomenon; high is defensible only for the narrower claim 'the IRS issued this warning.'
Not resolved — preserved on the record.
Objection (medium): Alternative explanation not considered: this is a routine, recurring IRS awareness reminder (fake charities are a perennial Dirty Dozen / disaster-season item), not a report of a detected new scheme. Publishing it as an alert implies an observed live campaign. The harvest cannot distinguish 'IRS is reminding people of an evergreen risk' from 'IRS has detected a current surge' — and the draft does not say which.
Not resolved — preserved on the record.
Objection (medium): Option sets are contaminated with unsupported targeting. callout_options offers 'ATTENTION: PARENTS' and 'ATTENTION: VETERANS' — neither group appears anywhere in the harvest. directive_options offers 'Send this to any commissioner you know' / 'fundraisers in your life' — 'commissioner' is a stray artifact of Commissioner Werfel's title, and fundraisers are described in the source as the people being asked questions, not as targets. Any selection from these lists would manufacture a targeting claim the evidence does not carry.
Resolved: Strike 'PARENTS', 'VETERANS', 'commissioner' and 'fundraiser' entries from the option sets; retain only 'EVERYONE', 'DISASTER APPEAL DONORS' and the selected seniors callout, all of which trace to the row. Keep share_directive null.
Objection (low): targeting_dropped rationale is factually wrong as written: 'phone — no harvested source describes this group.' Phone is a channel, not a group, and the source expressly describes spoofed caller ID and a 'real charity ... calling.' The watch_icons list correctly retains 'phone'. The stated reason contradicts the record and should be re-worded (e.g. 'no group tied to the phone channel is named') so the audit trail is accurate.
Resolved: Re-word the dropped-targeting note to state that the phone channel is evidenced (spoofed caller ID) but no audience group is tied to it, so no phone-based forwarding directive is issued.
Objection (low): Two small quantifier stretches: (a) source says criminals 'typically' seek money and personal information; the finding states flatly that 'these operations seek both'; (b) source advises 'Never work with charities that ask for donations by giving numbers from a gift card or by wiring money. That's a scam,' which the finding converts into the categorical 'any charity asking for donations by gift card numbers or by wiring money is a scam.' Both are close paraphrases but shed hedging in the direction of certainty.
Resolved: Restore the hedges: 'the IRS says these operations typically seek both money and personal information' and 'the IRS says a charity asking for gift card numbers or a wire transfer is a scam.'
Objection (low): Omission: the advisory also directs readers encountering a suspicious charity to the FBI's Charity and Disaster Fraud resources. That is the only reporting channel the source offers, and it is dropped from the finding — while the FBI still appears in the disclaimer with no context, which reads oddly. A one-clause addition would make the advice actionable rather than purely defensive.
Resolved: Add one clause to the advice: the IRS also points people who encounter a suspicious charity to the FBI's Charity and Disaster Fraud resources.
Preserved dissent
ON THE RECORDI do not accept the 'high' confidence rating. A single, unquantified, two-year-old awareness advisory — containing no victim count, no loss figure, no named entity, no geography and no enforcement action — supports the statement 'the IRS has published this warning' at high confidence and supports the statement 'criminals are doing this now' at medium at best. The draft's own limitations paragraph concedes the evidentiary emptiness and then the confidence field ignores it. That is the classic error of grading the source's letterhead instead of the claim.
ON THE RECORDPublishing an Oct. 2023 advisory under an undated present-tense claim in a masthead called LIVEFRAUD ALERTS is, in my view, the most serious defect here. It is not a fabrication, but it borrows currency the record does not supply. The date belongs in the claim line, not buried in the finding.
ON THE RECORDThe directive_options list ('Send this to any commissioner you know') indicates the option generator is pulling nouns out of a quoted official's job title. Even though share_directive is null this time, that failure mode will produce a targeting claim out of thin air on some future run and should be treated as a process defect, not a cosmetic one.
The sources
Official sourceIRS: Beware of fake charities; check before donating
The IRS warned taxpayers to be wary of criminals soliciting donations while falsely posing as legitimate charities, describing contact by email, fake websites and spoofed caller ID, requests for gift card numbers or wire transfers, and frequent targeting of seniors and groups with limited English proficiency.
Authority: official. Retrieved 2026-08-25.
Limitation: An advisory describing method and giving guidance; it reports no victim counts, no dollar losses, no geography and no specific fraudulent charity names.
Open the original source →Other checks
Approved by ihubglobalhq on 2026-08-26, after review of the alert and its sources.
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