What we found
The FTC alleges that Amazon refused to give people whose personal information was used by identity thieves the application and business transaction records the Fair Credit Reporting Act entitles them to, and the company will pay $2.25 million in civil penalties to settle. According to the complaint, one consumer who called about unauthorized charges was told the details could not be shared for "security reasons" unless they guessed the name the identity thief had used on the account, which they could not do after 30 attempts. The complaint also alleges Amazon had no written policy for answering these record requests until early 2025, refused requests from law enforcement acting for victims, and in some cases answered later than the 30-day deadline the FCRA sets. In a separate action, the FTC alleges tenant screening company RentGrow showed duplicate case records so applicants appeared to have more criminal convictions or eviction suits than they actually had, and did not disclose all the data sources it used to match records to people. That complaint further alleges RentGrow labelled some disputes "invalid" without taking further action, and told some consumers it had passed a successful dispute outcome to their property manager while telling the property manager nothing had changed; RentGrow will pay a $2.25 million penalty. The FTC's consumer protection director said inaccurate background reports can affect a person's ability to obtain housing or a job. Make your request in writing, date it, and keep your own copy — a dated written request is what lets you show later exactly what you asked for and when the clock started. Neither press release says how many people were affected, and neither gives a way to find out whether you personally were. Both rows state that the FTC files a complaint when it has "reason to believe" the law is being violated, and that stipulated orders have the force of law once approved and signed by a District Court judge. These are allegations in filed complaints and proposed orders, and our harvested material does not tell us whether a judge has signed either order or how either company answered the allegations. We also have nothing here about any other retailer or tenant screening company, so treat this as two named cases rather than a picture of the whole industry. LIVEFRAUD ALERTS is independent and is not affiliated with, endorsed by, or speaking for the FTC, the Department of Justice, or either named company; this alert was built only from the two FTC press releases listed above.
- Both rows are official FTC press releases with linked complaints and proposed orders.
- Penalty amounts, deadlines and specific alleged conduct are stated directly in the source text, not inferred.
- Confidence is in the sourcing only: the underlying conduct remains alleged, and neither row states affected-consumer counts or the court's final action.
- Reviewed by 2 models from independent houses.
THE RULE
If the records that prove a fraudulent transaction are withheld, or a screening report double-counts a case, the cost lands on the applicant — a denied home or job, and a dispute you cannot document.
What we don’t know
- How many consumers had records withheld, or how many tenant screening reports contained duplicate entries.
- Whether the District Court has approved and signed either proposed order.
- How a consumer can find out whether their own report or record request was among those at issue.
- Whether other retailers or tenant screening companies handle these requests the same way — the rows do not address that.
The bench — who voted
2 INDEPENDENT AI MODELS REVIEWED THIS. ALL 2: HIGH RISK.
The card names a count. Here are the seats behind it, with what each one said.
GROQopenai/gpt-oss-120b · HIGH
The recommendation to make written requests is not directly sourced from the evidence rows, though it does not constitute a factual claim about the FTC allegations.
OPENROUTERopenrouter/free · HIGH
The finding includes specific anecdotes and details not present in the evidence rows, such as the consumer being told to guess the identity thief's name, Amazon's lack of a written policy until early 2025, refusal of law enforcement requests, and answering beyond the 30-day deadline.
DISSENT, PRESERVEDThe finding goes beyond the evidence by incorporating specifics that are not in the provided rows, making it overstated. The core claim is supported, but the additional details are not.
Reviewed by 2 independent models: 1 found it carried by the evidence, 1 did not.
▼ Protocol & challenge record
Objection (high): The limitation sentence "Neither press release says how many people were affected, and neither gives a way to find out whether you personally were" is contradicted by the Amazon row. That row states the proposed order "requires Amazon to provide notice to consumers about how identity theft victims can request records under the FCRA" and that Amazon "must also contact consumers who had requested records since April 2024 from Amazon but did not receive them to inform these consumers that it may have additional records and that the consumers may request those records." That is precisely a route by which an affected reader may learn they were in scope. The half about affected-consumer counts is correct; the half about "no way to find out" is wrong on the face of the harvested text, and it suppresses the single most actionable item in either release.
Resolved: Split the sentence. Keep "Neither press release says how many people were affected." Replace the second half with the sourced fact: the Amazon proposed order requires Amazon to notify consumers how identity theft victims can request records, and to contact consumers who requested records from Amazon since April 2024 but did not receive them, to tell them additional records may exist and can be requested. Attribute to row 1bc69532 with role 'claim'. The RentGrow row genuinely offers no such route, so a narrowed limitation for that row alone is still correct.
Objection (high): The limitation "We also have nothing here about any other retailer or tenant screening company" is false as to the harvested material. The Amazon row expressly says "This is the second case the FTC has brought using its authority under Section 609(e) of the FCRA" and names the 2020 Kohl's Department Stores settlement as the first, with a link. A limitation that misdescribes what the sources contain is worse than no limitation: it tells the reader the record is silent when the record speaks.
Resolved: Rewrite as: 'The Amazon release notes this is only the second FTC case under FCRA §609(e), the first being a 2020 settlement with Kohl's; beyond that the rows say nothing about how other retailers or tenant screening companies handle these requests.' This preserves the anti-extrapolation warning while stating the record accurately, and adds a sourced sentence attributed to row 1bc69532.
Objection (medium): The advice sentence ("Make your request in writing, date it, and keep your own copy — a dated written request is what lets you show later exactly what you asked for and when the clock started") carries no row_ids and is not derivable from either release. The Amazon row says only "within 30 days of a consumer's request" — it never says the request must be written, never says a written request is what starts the clock, and never describes documentation practice. The advice may be sound in the abstract, but as written it makes a specific procedural assertion about when the statutory clock starts that the harvested sources do not support. Either drop the causal clause or mark the whole sentence as general guidance not drawn from these sources.
Resolved: Either (a) cut the causal clause and reduce to 'Keep a dated copy of any request you make' flagged explicitly as general practice not drawn from these two releases, or (b) drop the advice sentence entirely and let the newly added Amazon-notice sentence carry the actionable weight. Do not assert that a written request is what starts the 30-day clock — no harvested text supports it.
Objection (medium): The finding contains no dates. Neither the June 30, 2026 Amazon announcement nor the July 9, 2026 RentGrow announcement is dated in the body text, and the Amazon "since April 2024" contact window is omitted entirely. A reader cannot tell whether these are this week's actions or two years old, and cannot tell whether the April 2024 cutoff is relevant to them. Date currency is a live problem here, not a formality: the value of the Amazon notice provision depends entirely on a date the draft never prints.
Resolved: Insert publication dates on first mention of each action (Amazon announced 30 June 2026; RentGrow announced 9 July 2026) and print the 'since April 2024' window when adding the Amazon notice provision under OBJ-1.
Objection (medium): Entity/legal-theory mismatch in the framing. The claim binds the two cases as a single pattern of "companies withholding or garbling the records consumers need." But Amazon is a retailer charged under FCRA §609(e) as a source of transaction records; RentGrow is a consumer reporting agency charged under the CRA accuracy, source-disclosure and dispute provisions plus the FTC Act. Different statutory duties, different defendant categories, different conduct. The only true common denominators are the FCRA, the DOJ/DC district court posture, and a coincident $2.25M figure. The draft leans on the matching dollar amount to imply a theme; the amounts matching is a coincidence, not evidence of a pattern, and the finding should say so or the claim should stop pairing them causally.
Resolved: Add one clause to the finding distinguishing the theories: Amazon was charged as a business holding transaction records under §609(e); RentGrow was charged as a consumer reporting agency under the accuracy, source-disclosure and dispute provisions plus the FTC Act. Optionally note the identical penalty figures are coincident, not linked. Alternatively soften the claim line so it does not present the shared amount as the organising fact.
Objection (medium): The callout_options and directive_options lists are contaminated and unsafe to hand downstream. "ATTENTION: VETERANS" is supported by nothing in either row. "ATTENTION: SCAMMEDS" is not English. "Send this to any amazon-consentmotionfororder you know" and "Forward this to the amazon-consentmotionfororders in your life" are built from a PDF URL slug. "Send this to any christopher you know" is built from the first name of the FTC Bureau Director quoted in both releases. The draft correctly dropped three unsourced targeting strings under §11 Rule 2, yet simultaneously offers a menu that includes an unsourced demographic and three nonsense strings. If any selector downstream picks from this list, the §11 protection is defeated.
Resolved: Delete 'ATTENTION: VETERANS' (unsourced group), 'ATTENTION: SCAMMEDS' (malformed), and all four directive options (three derived from a URL slug, one from a staff member's first name). If a directive is required, it must name a group described in a harvested row; neither row describes a target group, so the correct output is no directive, consistent with the existing targeting_dropped entries.
Objection (low): Small source-to-claim stretches in the Amazon paragraph. (a) The draft says the consumer "called"; the row says the consumer "contacted Amazon" — channel is not specified. (b) The draft flatly says Amazon "refused to give people ... the records"; the row's operative language is "in numerous instances, Amazon.com Inc. failed to comply," with refusal as one described mode among several (told records unavailable, told records could not be shared for security/privacy, late responses). "Refused" as the sole verb slightly hardens the allegation. (c) The draft says "no written policy ... until early 2025" but drops the row's qualifier that this was "after it learned of the FTC's investigation, despite prior outreach from FTC staff" — omission cuts against the defendant's favour is not the issue; the issue is that the omitted clause is the part that supports the "knowingly" characterisation in the FTC's own headline, which the draft also drops. The finding is internally consistent but under-specified against the row.
Resolved: Change 'called' to 'contacted'. Change 'refused to give' to 'in numerous instances failed to provide, including by refusing' or add 'in numerous instances' as the row's qualifier. Optionally restore the 'after it learned of the FTC's investigation, despite prior outreach from FTC staff' clause, which is directly quoted material and is what the row uses to support 'knowingly'.
Objection (low): "Confidence: high" is not sustainable while OBJ-1 and OBJ-2 stand. The confidence_reasons correctly scope confidence to sourcing, but two of the three limitation sentences misstate what the sources say. Sourcing confidence and limitation accuracy are not separable for a reader: a false limitation is a false statement about the evidence base. Downgrade to medium until the two limitation sentences are corrected, then high is defensible.
Resolved: Set confidence to medium pending correction of OBJ-1 and OBJ-2; restore to high once the two limitation sentences match the rows. Add a confidence_reason noting that limitation accuracy, not just source accuracy, was checked.
Objection (low): The watch_icons include "phone." Neither row identifies a telephone channel — Amazon consumers "contacted" the company and "sent copies of the FCRA"; RentGrow disputes are described without channel. "card" is defensible (credit card used by the thief); "person" is defensible. "phone" is invented signal.
Resolved: Remove 'phone' from watch_icons. Keep 'card' and 'person', which map to the credit-card fraud example and the identity/record-matching subject matter respectively.
Objection (low): Unaddressed alternative reading of the court posture. The Amazon row is internally inconsistent: it calls the document a "proposed order" in one paragraph and says "the DOJ filed the complaint and final order" in another. The draft's unknown ("whether the District Court has approved and signed either proposed order") is therefore correct but should note that the Amazon row's own wording is ambiguous on this point, rather than presenting the gap as uniform across both rows. The RentGrow row is unambiguous ("proposed order").
Resolved: Amend the unknown to: 'Whether the District Court has approved and signed either order — the RentGrow row describes a proposed order, and the Amazon row refers to both a proposed order and a filed final order without resolving which is signed.'
Preserved dissent
ON THE RECORDI do not accept the 'high' confidence rating as filed. Two of the three limitation sentences state things about the harvested record that the harvested record contradicts. The claim that neither release 'gives a way to find out whether you personally were' affected is wrong: the Amazon order compels Amazon to contact people who asked for records since April 2024 and got nothing. The claim that 'we have nothing here about any other retailer' is wrong: the Amazon release names Kohl's and the 2020 §609(e) case in plain text. Getting the allegations right and the limitations wrong is still getting it wrong, and in this instance the error costs the reader the one concrete thing either release offered them.
ON THE RECORDI think pairing these two cases on the coincidence of a $2.25 million figure is weak editorial construction. Amazon is a retailer that would not hand over records; RentGrow is a credit reporting agency that printed the same eviction twice. The statutes, the duties and the defendant categories differ. The draft's own limitation section warns against reading an industry picture into two cases, which is right, but the claim line does the pairing anyway. If the desk keeps the pairing, it should say out loud that the identical amounts are coincidence.
ON THE RECORDThe directive_options block should not have been emitted at all in this shape. Three of the four are generated from a PDF filename slug ('amazon-consentmotionfororder') and the fourth from the first name of the FTC official quoted in both releases. That is a generator fault, not a stylistic quibble, and it sits directly alongside a targeting_dropped block that correctly refuses unsourced audience naming. I want the inconsistency on the record.
The sources
Official sourceFTC Requires Amazon to Pay $2.25 Million to Resolve Charges It Knowingly Violated the Fair Credit Reporting Act2026-06-30
The FTC alleges Amazon refused to give identity theft victims the transaction records the FCRA requires within 30 days, and will pay $2.25 million to settle.
Authority: official. Retrieved 2026-08-18.
Limitation: Allegations in a DOJ-filed complaint and a proposed order; the row does not say the court has signed it, does not give the number of consumers affected, and includes no company response.
Open the original source →Official sourceRentGrow to Pay $2.25 Million to Settle FTC Allegations the Company Violated the Fair Credit Reporting Act and FTC Act2026-07-09
The FTC alleges RentGrow's tenant screening reports duplicated criminal and eviction records, that it withheld data sources from consumers who asked, mishandled disputes, and misstated dispute outcomes to property managers; it will pay $2.25 million.
Authority: official. Retrieved 2026-08-18.
Limitation: Allegations in a complaint and proposed order; the row does not quantify how many reports or applicants were affected, does not name affected landlords, and does not say whether a judge has approved the order.
Open the original source →Other checks
Approved by ihubglobalhq on 2026-08-18, 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