What we found
The Federal Trade Commission says it is sending payments totalling more than $23.8 million to drivers harmed by Grubhub's deceptive earnings claims and to diners harmed by the company's conduct, including blocked accounts and funds. According to the agency, checks or PayPal payments are going to 640,038 affected consumers, with most receiving a check in the mail, checks to be cashed within 90 days and PayPal payments redeemed within 30 days. The FTC states plainly that the Commission never requires people to pay money or provide account information to get a payment, and that it will never demand money, make threats, tell you to transfer money, or promise you a prize. The only consumer contact point named in the announcement is the refund administrator, Analytics Consulting LLC, on 1-888-446-4992, alongside the FTC's own website for questions about the redress process. A published refund with a named administrator, a fixed dollar figure and a ticking redemption clock is exactly the kind of announcement an impostor can read aloud back to you. If anyone contacts you about this money and asks for a fee, a transfer, or your bank or card details to release it, stop and end the contact — a real refund never needs either. We have not harvested any report of someone impersonating the FTC or the refund administrator in connection with these payments, so we are not claiming that is happening. Nothing in our harvested material tells us how contact would be attempted, whether by call, message or email, or whether any consumer has lost money in relation to this redress. LIVEFRAUD ALERTS is an independent consumer awareness service and is not affiliated with the Federal Trade Commission, the Illinois Attorney General, Grubhub, Analytics Consulting LLC or PayPal, which are named here only as they appear in the official announcement.
- The refund programme, the amount, the recipient count and the payment deadlines all come directly from a single official FTC press release.
- No harvested row reports any actual impersonation, contact attempt or consumer loss tied to this refund, so the risk framing rests on the announcement's own warning rather than observed activity.
- We have only one source and no detector or community material to corroborate or extend it.
- Reviewed by 2 models, 1 from independent houses.
THE RULE
Named administrator, known amount and a 30-to-90-day clock give a caller everything needed to sound official — the tell remains any request for money or account details.
What we don’t know
- Whether anyone has been contacted by someone posing as the FTC or the refund administrator about these payments.
- What contact channels any such approach would use.
- Whether any recipient has lost money or handed over account details in connection with this redress.
- What happens to payments that are not cashed or redeemed within the stated windows.
The bench — who voted
2 INDEPENDENT AI MODELS REVIEWED THIS. ALL 2: MODERATE RISK.
The card names a count. Here are the seats behind it, with what each one said.
ANTHROPICclaude-opus-5 · MODERATE
The single evidence row supports only the total ($23.8m), the recipient count (640,038), the check/PayPal method and the FTC's standing 'never pay, never give account details' warning. The finding adds several specifics with no evidentiary backing here: the 90-day check and 30-day PayPal redemption windows, the refund administrator's name (Analytics Consulting LLC), the phone number 1-888-446-4992, the expanded warning language ('threats... transfer money... promise you a prize'), and the reference to the Illinois Attorney General. These may well be in the underlying release, but the panel was given no row that says so.
DISSENT, PRESERVEDThe absence-of-evidence disclosures here are unusually candid and should not be penalised — the drafters state plainly that no impersonation report was harvested and that contact channels and losses are unknown. My objection is to the claim line and the unsourced operational details, not to the framing discipline in the body.
DISSENT, PRESERVEDThere is a countervailing harm worth recording: alerts that attach suspicion to a genuine government refund can cause recipients to discard legitimate mail or leave checks uncashed past the redemption window. Any published version should say explicitly that receiving a check in the mail is the expected, legitimate route.
GROQopenai/gpt-oss-120b · MODERATE
Raised no objection; read the evidence as holds.
Reviewed by 2 independent models: 1 found it carried by the evidence, 1 did not.
▼ Protocol & challenge record
Objection (high): The claim line asserts as established fact that "the payment window itself is what impostors work with." No harvested row reports a single impersonation attempt, contact, or loss tied to this redress — the finding itself concedes this twice. The claim is a hypothesis dressed as a finding. Everything downstream (risk_line, watch_icons, the whole LIVEFRAUD ALERTS framing) inherits an assertion the evidence does not carry. The disclaimers deeper in the body do not cure a headline claim that states the speculative part in the indicative mood.
Not resolved — preserved on the record.
Objection (medium): Entity/mechanism mismatch: "The FTC is mailing more than $23.8 million" overstates the source. The release says the FTC is sending checks OR PayPal payments and that "most" consumers will receive a check in the mail. Some meaningful share of the $23.8m is not being mailed at all. Say "sending," not "mailing."
Resolved: Trivial wording fix available: change "mailing" to "sending" in the claim and note that most, not all, recipients get a mailed check. The finding sentence already says this correctly; only the claim line needs correcting.
Objection (medium): Source-to-claim stretch on the safety rule. The release says "The Commission never requires people to pay money or provide account information to get a payment" — a statement about the FTC. The claim generalises to "no legitimate refund asks you to pay or hand over account details." That is a broader universal than the source supports, and it is factually shaky in this very case: redeeming a PayPal payment requires the recipient to log in to a PayPal account. An impostor can exploit exactly that gap ("just confirm your PayPal login to release the payment"), and the alert as written gives the reader a rule that collides with the real redemption flow instead of warning about that specific seam.
Not resolved — preserved on the record.
Objection (medium): Date currency. The source is dated 2026-08-12. The 90-day check window runs from the date printed on each check, not from the announcement, and the 30-day PayPal window is shorter still. If this alert publishes more than a few weeks after 2026-08-12, the "ticking clock" hook is stale or expired and the advice becomes actively unhelpful. There is no "as of" date anywhere in the finding and no statement that the clock starts from the check date rather than the press release. The risk_line's flat "30-to-90-day clock" compounds this by implying a single uniform deadline from a single start point.
Not resolved — preserved on the record.
Objection (high): Alternative explanation / net-harm inversion, unaddressed. The best-documented failure mode around FTC redress mailings is not impostors — it is legitimate recipients binning real checks because they assume a surprise government cheque is a scam. This piece attaches fraud suspicion to a genuine payment and gives the reader no positive verification path: it never tells them the check is real, never points to ftc.gov/refunds as the way to confirm, and never says "if a check arrives in the mail unprompted, that is what the real thing looks like." As drafted, the most likely reader harm is an uncashed valid check, not a prevented fraud. The alert should carry the verification half or it should not run.
Not resolved — preserved on the record.
Objection (high): The callout and directive option sets are defective. "ATTENTION: SCAMMEDS" is not a word. "ATTENTION: ADMINISTRATORS," "Send this to any administrator you know" and "Forward this to the administrators in your life" appear to be generated by mistaking the refund administrator (a corporate claims processor) for a population of readers — there is no such audience and no evidence any such group is affected. "ATTENTION: EVERYONE" is the opposite failure, unbounded targeting. Only "ATTENTION: FOOD DELIVERY DRIVERS" and the first two directives are defensible, and the first directive ("a driver who was blocked from their account or their funds") mixes the two harm classes — the release attributes blocked accounts and funds to diners, not drivers.
Not resolved — preserved on the record.
Objection (low): targeting_dropped contains a broken string: "DIRECTIVE NOT IN EVIDENCE: expecting — no harvested source describes this group." "expecting" is not a group; the record of what was dropped and why is unreadable.
Resolved: Regenerate or delete the targeting_dropped string; the underlying rule application (no evidence any named group is being targeted) is correct even though the note is garbled.
Objection (medium): watch_icons [phone, bank, link] contradict the finding's own limitation that "Nothing in our harvested material tells us how contact would be attempted, whether by call, message or email." Three channel icons assert a channel profile the piece explicitly disclaims knowing. At most a generic icon is warranted; "link" in particular has zero footing — no URL, SMS or email vector appears anywhere in the evidence.
Not resolved — preserved on the record.
Objection (low): "The only consumer contact point named in the announcement is the refund administrator" is not quite right — the release also names the FTC Office of Public Affairs on 202-326-2180. It is labelled a media contact, so the sentence is nearly true, but "the only contact point named" is falsifiable on its face and an alert about impostor phone numbers should be precise about which numbers appear in the official text.
Resolved: Amend to "the only consumer contact point named" is already qualified by "consumer" in the finding sentence — adding "(a separate number is listed for media only)" closes it fully. Low cost, no evidentiary problem.
Objection (low): "asks for a fee, a transfer, or your bank or card details ... a real refund never needs either" — "either" governs three items. Small, but it is the single most important sentence in the piece and it does not parse.
Resolved: Rewrite as "a real refund needs none of these."
Objection (low): Confidence is recorded as "moderate" for the whole item, which blends two very different confidence levels: near-certain on the factual redress details (direct from an official primary source) and near-zero on the impostor-risk framing (no observations at all). One label conceals that the weakest component is the one carried in the headline claim.
Not resolved — preserved on the record.
Preserved dissent
ON THE RECORDI do not think this should run in its present form. There is no fraud here. There is one official press release about a legitimate government refund, and the piece manufactures a threat frame out of it — "the payment window itself is what impostors work with" — with nothing behind that sentence but the FTC's standard boilerplate warning, which appears at the foot of every FTC release ever published. If boilerplate is sufficient to generate an alert, then every FTC release generates an alert, and the alert carries no information.
ON THE RECORDThe internal disclaimers are honest and I credit them, but they sit below a claim line that states the unsupported part flatly. Readers take the claim. Three limitation sentences cannot unsay a headline.
ON THE RECORDMy substantive worry is that this alert is net-harmful. The known failure mode with FTC redress mailings is people throwing away real checks because a surprise government cheque smells like a scam. This piece pours suspicion onto a genuine payment, attaches phone/bank/link warning icons to it, and never once tells the reader that the check is real, that it will arrive unprompted, and that ftc.gov/refunds is where to confirm it. On the evidence available, the most probable reader outcome is a binned valid check, not a foiled impostor. If the piece runs, the verification half is not optional.
ON THE RECORDThe callout and directive options are not a formatting nit. "ATTENTION: SCAMMEDS" is gibberish and the four "administrator" variants come from mistaking a claims-processing vendor for an audience segment. That is the same class of error the targeting rule exists to catch, and it got through into the option set.
The sources
Official sourceFTC Sends More than $23.8 Million to Drivers and Diners Harmed by Grubhub’s Deceptive Advertising Claims and Other Unlawful Conduct2026-08-12
The FTC says it is sending more than $23.8 million to 640,038 Grubhub drivers and diners by check or PayPal, and states the Commission never requires people to pay money or provide account information to get a payment.
Authority: official. Retrieved 2026-08-24.
Limitation: The release describes the redress programme and the agency's standing warning; it does not report any scam, impersonation attempt or consumer loss connected to these payments.
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 Sends More than $23.8 Million to Drivers and Diners Harmed by Grubhub’s Dece".
- ✓ All 4 material sentence(s) map to FTC.
- ✗ 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 2026-08-24.
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