Domain Atlas / Lending & credit collections AI
TransUnion's OFAC Name Screen & the people who could not sue
Explore this deployment in the PAN Lab ↗
In the PAN Lab, the readouts of this case's model organization carry a shaded evidence band whose width follows the least-established class among the modeling inputs the readings rest on.
The least-established input behind this case's model organization's readings is an assumption, not a measurement. Evidence base: 1 assumed · 12 published baseline.
From 2002 TransUnion LLC, one of three nationwide consumer reporting agencies, sold subscribers an add-on to the ordinary credit report — marketed and litigated variously as 'OFAC Advisor', 'OFAC Name Screen' and, in TransUnion's own filings with the Securities and Exchange Commission, 'the OFAC Alert service'. On a subscriber credit pull carrying the append, TransUnion passed the consumer's first and last name, and nothing else, to third-party software and data held by a vendor, Accuity, Inc., which compared it against the U.S. Treasury Department's Specially Designated Nationals and Blocked Persons list; a match was written into the 'SPECIAL MESSAGES' section on the front page of the report sold to the subscriber. The Ninth Circuit found that TransUnion introduced the product on the belief that it was exempt from the Fair Credit Reporting Act because the data sat in the vendor's file rather than its own database, that on that basis 'TransUnion did not follow its normal procedures to ensure accuracy', and that it adopted a policy of not disclosing OFAC matches to consumers who requested their own reports. It summarised the result as name-only searches 'for more than a decade, resulting in thousands of false positives and not a single known actual match identified' — an evidentiary absence in one case over one class window rather than a proven zero. On 27 February 2011 Sergio Ramirez was refused the sale of a car at a Dublin, California dealership after a credit check returned an OFAC alert and a salesman told him his name was on a 'terrorist list'; his wife bought the car in her own name.[4]
What happened
TransUnion is one of three nationwide consumer reporting agencies. From 2002 it sold subscribers an add-on to the ordinary credit report, marketed and litigated variously as "OFAC Advisor", "OFAC Name Screen" and, in its own filings with the Securities and Exchange Commission, "the OFAC Alert service". When a bank, a landlord or a car dealership pulled a consumer's report with the append attached, TransUnion took the consumer's first and last name and passed only that name to third-party software and data held by a vendor, Accuity, Inc., which compared it against the Treasury Department's Specially Designated Nationals and Blocked Persons list — the list of people barred from transacting business in the United States. If the names matched, the front page of the report carried an alert in its special-messages block, printed as "SPECIAL MESSAGES".
The comparison used two fields. No date of birth, no middle initial, no Social Security number, no citizenship and no address entered it at any point. Before November 2010 it fired on names that were, in the Ninth Circuit's words, "either identical or similar" — the court's worked example is that "Cortez" would match "Cortes". There was no threshold to tune, no probability and no calibration surface. It was a string equality test against a watchlist.
Two facts about that design are what make this case unusual, and both are findings rather than characterisations. The first is that the disambiguating data existed on both sides. TransUnion held the consumer's date of birth and Social Security number in its own database, CRONUS, and the Third Circuit recorded that a creditor had to supply at least a name AND an address to retrieve a file from it — a richer key than the one TransUnion itself sent to Accuity, "even though Trans Union may have more information about the person who is the subject of the inquiry." The sanctions records on the other side carried the listed persons' names, dates of birth and passport numbers. TransUnion reprinted exactly those fields in the letter it mailed the consumer. And in between: "Trans Union does no other comparison or due diligence with the data it receives from Accuity to attempt to match it to the consumer whose credit report is being furnished." The Ninth Circuit added the comparison that makes it starkest: for tax liens and bankruptcy judgments TransUnion used at least one identifier besides the name, and "OFAC information was the only consumer-report data that TransUnion collected using name alone."
The second is why the data sat where it sat. TransUnion introduced the add-on in 2002 on the belief that it was exempt from the Fair Credit Reporting Act, specifically because the sanctions data was not stored in its own database but in a separate file and software supplied by Accuity. On that basis, the Ninth Circuit found, "TransUnion did not follow its normal procedures to ensure accuracy", and it adopted a policy of not disclosing OFAC matches to consumers who requested their own reports. Architecture and legal theory were the same move.
The corrective loops were closed at both ends. The consumer's own copy of the report did not show the alert, so the only person with the knowledge and the motive to correct the label could not see it; TransUnion's witnesses acknowledged that the personal credit reports it gives consumers never show any information or alerts from the product it provides to creditors. And the statutory dispute channel — the Fair Credit Reporting Act's own error-correction mechanism — was switched off for this data class by policy: "once Trans Union receives the OFAC information it does not check or confirm its accuracy; in fact, Trans Union has a policy of never reinvestigating disputes involving OFAC alerts." Its call centre told both named plaintiffs there was no alert on their report while an alert sat on the version being sold. What TransUnion did do is the sharpest number in the record: the district court found that it "removed the OFAC Alert of each class member who contacted Trans Union following receipt of the OFAC letter", and cited that against the company's own argument that a correction was technically infeasible. Correction conditional on discovery was total; discovery was suppressed by design.
Sandra Cortez sued in 2005 after a Colorado car dealership confronted her with an alert; a jury awarded her $50,000 in actual and $750,000 in punitive damages, remitted to $100,000, and the jury handwrote on the verdict form a plea that Trans Union "completely revam[p]" its business practices. In August 2010 the Third Circuit affirmed, holding that OFAC alerts are part of the consumer report and subject to the statute's maximum-possible-accuracy duty: "We do not believe that Congress intended to allow credit reporting companies to escape the disclosure requirement... by simply contracting with a third party to store and maintain information that would otherwise clearly be part of the consumer's file." It rejected the subscriber addendum as a defence — "We are not persuaded that Trans Union's private contractual arrangements with its clients can alter the application of federal law" — and called the failure "to take the utmost care in ensuring the information's accuracy — at the very least, comparing birth dates when they are available" reprehensible.
Two months later, officials at Treasury's own OFAC wrote to TransUnion. They said they continued to hear from TransUnion's customers and from individual consumers adversely affected by false alerts, warned that a product "that does not include rudimentary checks to avoid false positive reporting can create more confusion than clarity and cause harm to innocent consumers", and said they were "particularly worried" by OFAC alerts being "disseminated broadly in conjunction with credit reports." The letter named the defect precisely and carried no enforcement power over TransUnion.
The Ninth Circuit found that after all this the company "made surprisingly few changes". In November 2010 it re-worded the alert from "match" to "potential match" and tightened the matcher to require an exact first-and-last-name match, which the court records as cutting the false-positive rate from about five per cent to about half a per cent. It requested further software enhancements from Accuity; those were not implemented until 2013, and it "continued to use only first and last names to identify OFAC matches until 2013". Justice Thomas, dissenting, summarised the response: "It did not begin comparing birth dates. Or middle initials. Or citizenship. In fact, TransUnion did not compare any new piece of information."
Between 1 January and 26 July 2011 TransUnion used a two-mailing practice. Envelope one carried the consumer's credit report with the OFAC alert redacted, plus the statutory summary of rights. Envelope two, arriving a day later, was the "OFAC Letter": it named the potential match and reprinted the matched sanctions records with their names, dates of birth and passport information, gave no dispute instructions, omitted the summary of rights, and never stated that the alert appeared on the version of the report sold to third parties. TransUnion stopped that practice in July 2011.
On 27 February 2011, inside that window, Sergio Ramirez was refused the sale of a car at a Dublin, California dealership with his wife and father-in-law present. A salesman told him Nissan would not sell him the car because his name was on a "terrorist list"; his wife bought the car in her own name. He consulted a lawyer and cancelled a planned family trip to Mexico. He filed a class action in February 2012, and in July 2014 a class of 8,185 was certified: everyone who had received an OFAC Letter in the seven-month window.
In March 2017 the district court denied summary judgment, reasoning that a jury could find the alert misleading "given that the evidence supports a finding that none of the consumers flagged as a potential match were in fact a match; in other words, a jury could find that if Trans Union had used more information than just a matching name to flag a consumer — such as a matching birth date — none of the class members would be even a potential match." It pointed to two facts. Three TransUnion witnesses testified there was no evidence that any consumer whose file carried an alert was in fact on the list. And an Experian report on Ramirez in the same period read "NAME DOES NOT MATCH OFAC/PLC LIST" — screen output for the same person, the same list, the same period, a different bureau, no alert.
On 21 June 2017 the jury found for the class on all three claims, awarding $984.22 in statutory and $6,353.08 in punitive damages per class member, about $60 million in total and reported at the time as the largest verdict under the statute to that date. The Ninth Circuit affirmed liability and standing in February 2020 while holding the punitive award constitutionally excessive, reducing it to $3,936.88 per member. It summarised the decade in one sentence: name-only OFAC searches "for more than a decade, resulting in thousands of false positives and not a single known actual match identified."
Then the case changed shape. On 25 June 2021 the Supreme Court held 5-4 that "No concrete harm, no standing." The 1,853 class members whose reports were provided to third-party businesses had suffered a harm bearing a close relationship to defamation and could sue; the 6,332 whose reports were never disseminated could not, because "the mere existence of inaccurate information, absent dissemination, traditionally has not provided the basis for a lawsuit in American courts" — the plaintiffs' harm, the Court said, was "roughly the same, legally speaking, as if someone wrote a defamatory letter and then stored it in her desk drawer." Footnote 5 is the sentence that governs how this case may be summarised: "For purposes of this case, the parties have assumed that TransUnion violated the statute even with respect to those plaintiffs whose OFAC alerts were never disseminated to third-party businesses... We take no position on that issue." The Court decided who could sue. It did not decide whether the product was accurate or lawful, and no court ever enjoined it or ordered a design change.
The dissenters attacked the population mismatch directly. Justice Thomas, joined by three colleagues: "in a 7-month period, it is undisputed that nearly 25 percent of the class had false OFAC-flags sent to potential creditors. Twenty-five percent over just a 7-month period seems, to me, [a degree of risk sufficient to meet the concreteness requirement]. If 25 percent is insufficient, then, pray tell, what percentage is?" The majority answered that the plaintiffs bore the burden of proving their reports were actually sent, and that many of the 6,332 "would first learn that they were [injured] when they received a check compensating them for their supposed [injury]." The United States, filing as amicus supporting neither party with counsel from the Consumer Financial Protection Bureau, had argued that all class members suffered an injury in fact where TransUnion "willfully violated 15 U.S.C. 1681e(b) by producing consumer reports that erroneously designated the class members as individuals who are barred from engaging in transactions in the United States."
On remand the case settled rather than producing a merits judgment. TransUnion's Form 10-K for fiscal year 2022 records it in one sentence: "On December 19, 2022, the court entered final approval of the class settlement and we paid the settlement amount to the plaintiffs on January 20, 2023, resulting in a full resolution of this matter." The filing does not state an amount; legal trade press reports a $9 million fund, $4.2 million in class-counsel fees, a $75,000 service award and an estimated recovery above $2,000 per claimant, and gives 15 December 2022 as the final-approval hearing date, four days before the date the 10-K gives for the order. Taking judicial notice of the federal record, the Illinois Appellate Court recorded who was covered: "the 1,853 class members determined to have standing, as well as an additional 147 class members who demonstrated standing pursuant to a claims process." Two thousand of the original 8,185.
On 23 January 2023 the district court dismissed the remaining class members' claims without prejudice. The same day, one of them — Jose Luis Ramirez Arrizon Jr. — filed a putative class action in the Circuit Court of Cook County, Illinois, raising the same disclosure and summary-of-rights claims in a forum without the Article III standing rule federal courts apply. The circuit court dismissed it as time-barred and the Illinois Appellate Court affirmed on 31 March 2025, declining equitable tolling and noting that "the Ramirez majority took the position that [concrete harm] has always been required for standing." The route out of federal court closed.
What changed to the product afterwards is only partly on the record, and the honest answer has three tiers. Verified: TransUnion's filings state that "as a result of a decision by the [Third Circuit] in 2010 (Cortez v. Trans Union LLC), we modified one of our add-on services", and the Ninth Circuit found the modifications were the November 2010 wording change and the exact-match tightening, with the Accuity enhancements landing in 2013. Alleged only, in a 2020 complaint that was voluntarily dismissed with prejudice in March 2022 with no class certified and nothing adjudicated: that TransUnion represented during the Ramirez litigation that it gained the ability to consider dates of birth in 2013, and that it continued to place alerts where the sanctions record's birth date visibly differed. Not established by anything verified: what the matcher compares today. A TransUnion master agreement dated March 2015, filed with the Securities and Exchange Commission in 2020, still describes the screened name as the one "supplied by Subscriber to TransUnion on input and not as may be found on TransUnion's database(s)", and still makes the subscriber "solely responsible" for any required action while barring it from taking adverse action "based, in whole or in part, on TransUnion's OFAC Name Screen services." TransUnion's Form 10-K for fiscal year 2025 contains no mention of OFAC or Ramirez at all.
One separate matter is named here only to keep it separate. In October 2023 the Federal Trade Commission and the Consumer Financial Protection Bureau announced a $15 million settlement with TransUnion Rental Screening Solutions, Inc. and Trans Union LLC over eviction and criminal-record accuracy in tenant screening reports. It is a different product and has nothing to do with sanctions name screening. It is the only federal regulatory action against this operator on accuracy in the relevant period, which is exactly why the confusion is worth naming rather than leaving to a reader.
The sociotechnical reading
Every other lending case in this atlas turns on a model output, a threshold, a proxy or a disparate-impact statistic. This one has none of those. The entire decision is a name equality test against a terrorism and narcotics watchlist, and the harm is being labelled a potential terrorist inside a credit file. Four things follow from that, and the fourth is why the case is here.
The first is that this is a deliberate under-join rather than a data gap. The usual story about automated error is that the system did not have enough information. Here the information existed on both sides of the comparison, in the same transaction, in the same company. TransUnion held the consumer's date of birth and Social Security number in its own database and demanded a name and an address before it would let a creditor retrieve from it. The sanctions records carried the listed person's date of birth and passport number. TransUnion literally printed those fields and mailed them to the consumer. The pipeline was built not to put one against the other, and the district court's summary judgment reasoning is that a single one of those fields would have disposed of the whole class. When people ask what a governance failure looks like as distinct from a technical one, this is the cleanest available answer: nothing was missing, and nothing was joined.
The second is that the vendor boundary was simultaneously the architecture and the legal theory. The reason the sanctions data sat in Accuity's file rather than in TransUnion's own database is the same reason TransUnion concluded in 2002 that the product fell outside the Fair Credit Reporting Act, and therefore the same reason it did not apply its normal accuracy procedures to this data class. That is a single decision wearing two hats, and it held for eight years until a court took one of them off. It also shows why procurement is a civil-rights surface: everything a serious vendor gate asks for — a documented performance claim, a right to inspect, a timeline for a change — is absent from this relationship, and its absence is measurable. TransUnion could not confirm that a single alert it had sold was accurate. It asked Accuity for improvements in 2010 and got them in 2013.
The third is liability export, and its two halves came apart. Legally it failed twice: the Third Circuit held that private contractual arrangements cannot alter the application of federal law, and the district court rejected the argument that contractual human review discharged the statutory duty. Socially it worked. The contract told the subscriber it was solely responsible and must not act on the alert; the operator's own appellate framing is that the product "casts a wide initial net and then relies on a lender's human judgment." What that judgment looked like on the ground is the record's two documented encounters. A salesman refused a sale. A finance manager, in the predecessor case, phoned his regional finance director, went to the Treasury website himself, asked the consumer whether she had ever lived outside the United States and said he would have to check with the FBI. Neither had a procedure. Treasury publishes exactly the guidance that desk needed — a reviewer who does not believe the consumer is the listed person "should disregard the alert"; one who does "should call the OFAC Hotline to verify" — and it reached the finance desk through no channel this deployment operated. A design that puts its last line of defence in another company's untrained employee has not delegated the decision; it has abandoned it.
The fourth is the governance question this case exists to pose, and it is not about accuracy at all. The Supreme Court fixed the boundary of a cognisable harm at dissemination rather than at labelling. Consider where that line falls. Whether a business happens to pull your file inside an arbitrary seven-month window is a fact about the operator's customers, not about you; you did not cause it, you cannot influence it, and — because the alert was redacted from your own copy of the report — you cannot observe it. So the boundary between the people who could go to court and the people who could not was drawn at a step the operator controlled and the labelled person could not see. 8,185 people carried a label produced by the same defect. 1,853 could sue. On remand 147 more proved dissemination through a claims process and joined the settlement; 2,000 people were paid out of 8,185. The rest were dismissed without prejudice, and when one of them tried a state forum that has no concreteness requirement, the claim was held time-barred. The population a system harms and the population that can challenge it came apart, permanently, on a boundary the system itself drew.
It is worth stating plainly what the decision did not do, because the case is often summarised as if it settled the design. It did not. Footnote 5 expressly reserves whether the statute was violated as to the 6,332. Nothing after the remand produced a merits judgment. No court enjoined the product, no regulator has ever taken enforcement action over it, and TransUnion's standard contract still carried an OFAC Name Screen clause in 2015. The only final appellate liability holding against this practice remains Cortez, decided in 2010 on the facts of one consumer in Colorado, and the thing that changed the matcher was not a regulator or an audit or an internal review — it was a jury, and it took eight years to arrive.
And one last measurement, because it is the pair that describes this deployment better than any single number. Every consumer who complained had the alert removed: a correction rate of one hundred per cent, cited by the court against the company's own claim that correcting was infeasible. And the alert was invisible in the only document that would have told a consumer there was something to complain about. A safeguard that works perfectly whenever it is reached, behind a door the operator kept shut, is not a safeguard. It is a measurement of who was expected to find out.
The concepts used in this reading are defined in the Field Guide; the governance responses live in the Practice Library. The model organization for this case can be stress-tested in the PAN Lab.