Tiffany found 73.1 and 75.5 percent of its 2004 and 2005 eBay test buys counterfeit - and the court held generalized knowledge imposed no duty on eBay

The document is an appeals court opinion in Tiffany's trademark suit against eBay, held in the Caselaw Access Project collection at the path f3d/600/cases/0093-01.json. The research file behind this piece identifies it as the Second Circuit's decision of April 1, 2010, reported at 600 F.3d 93; the captured text carries no date, no court designation and no citation line of its own, so that identification is the file's and not the document's. What the capture does carry is a jeweler's own count of fakes bearing its name, and a rule that made the count legally beside the point.

The claim being tested against it was published sixteen years later. The Office of the United States Trade Representative released its 2025 Review of Notorious Markets for Counterfeiting and Piracy; the captured release carries no date in its text, and March 2026 is taken from the URL path of the capture. The release states that the list identifies "37 online markets and 32 physical markets" reported to engage in or facilitate substantial trademark counterfeiting or copyright piracy. It also carries a passage written in the first person about the tournament: with the United States co-hosting the FIFA World Cup, the release states, there is particular attention to "sales of counterfeit merchandise and illicit streaming of sports broadcasts", activities it describes as intellectual property theft that also harm consumers through substandard goods and malware. The captured passages attach no name and no office to those words. This piece therefore attributes them to the release, because the capture identifies no speaker.

That is the enforcement instrument: a document that names where counterfeiting is reported to happen, on what this desk reads as the premise that naming it moves the parties who host it. The captured passages do not claim to enumerate particular infringing listings.

Now the record. The opinion states that sometime before 2004 Tiffany became aware that counterfeit Tiffany merchandise was being sold on eBay's site, and that it ran two surveys, called Buying Programs, one in 2004 and one in 2005. Under those programs Tiffany bought items on eBay, then inspected and evaluated them. Tiffany found that 73.1 percent of the purported Tiffany goods bought in the 2004 program were counterfeit, and 75.5 percent of those bought in 2005. Both figures are Tiffany's own findings as the opinion reports them, not measurements made by a court. The captured opinion does not state how many items either program bought.

The same opinion records that the district court concluded the Buying Programs were "methodologically flawed and of questionable value". Separately, in a passage the capture does not connect to the surveys, the opinion states that eBay certainly had generalized knowledge that Tiffany products sold on its site were often counterfeit. The capture does not say the surveys produced that finding, and this piece does not claim they did.

Then the rule at the center of the captured opinion. Generalized knowledge, the court held, "is insufficient under the Inwood test to impose upon eBay an affirmative duty to remedy the problem." The standard the opinion sets out instead: "Some contemporary knowledge of which particular listings are infringing or will infringe in the future is necessary." On willful blindness, the opinion states the test as suspecting wrongdoing and then deliberately declining to investigate. The capture stops at the standards it states: it carries no judgment or disposition line, and this piece asserts none.

The opinion carries the cost allocation in the same document as the percentages. It describes the Verified Rights Owner program, a notice-and-takedown system eBay maintained for nearly a decade, including the period at issue, through which rights owners - Tiffany among them - report listings offering potentially infringing items so that eBay can remove them. The reporting is the rights owner's work. The opinion also records eBay spending of "as much as $20 million each year" on tools to promote trust and safety on its website. A reader who wants the counterfeit rate, the takedown burden and the platform's spending will find all three in the same document.

Analysis, and it is this desk's reading rather than anything a source states: the Notorious Markets List operates in precisely the evidentiary category the opinion holds insufficient. It publicises generalized knowledge of where counterfeiting occurs. The 2010 holding says generalized knowledge, however well documented, imposes no affirmative duty; only contemporary knowledge of particular listings does. The two documents count different things - one counts markets, the other counted items in a plaintiff's own test buys - and they concern different parties and different questions sixteen years apart. Reading the second against the first is a comparison, not a finding that either is wrong.

The test-buy method itself did not stop with Tiffany. The Government Accountability Office reported in GAO-18-216 that "20 of 47 items GAO purchased from third-party sellers on popular consumer websites were counterfeit". CBS News, reporting on the same audit, states that investigators bought 47 products from third-party sellers hosted on five sites - Amazon, Walmart, eBay, Sears Marketplace and Newegg - and that 20 were determined to be counterfeit by the companies holding the products' intellectual property rights. The determinations were the rights holders', not the auditors'.

The breakdown, per CBS News: all 13 Urban Decay cosmetics purchased across the five sites were determined counterfeit, as were 6 of 9 Yeti mugs and 1 of 10 UL-certified phone chargers. All 15 pairs of Nike Air Jordan shoes were found to be authentic. No source in this piece accuses any of the five platforms of knowing wrongdoing; the sellers were third parties.

What a reader can check next is the next review. USTR's list is an annual instrument, and the 2025 edition is the one captured here. This desk predicts that if USTR publishes a further Notorious Markets List on or before June 30, 2027, that list will again identify markets rather than enumerate the particular infringing listings the 2010 opinion requires before liability attaches to a marketplace.

The other thing to watch is the rule itself. Nothing in the captures used here shows a later court or Congress moving it - the record in this file stops at the opinion itself. While it stands, a brand can buy its own goods on a platform, find that between 73.1 and 75.5 percent of them are fake, publish the number, and still have shown only the generalized knowledge the captured opinion holds insufficient to impose an affirmative duty on the marketplace.