BrandShield published its 2026 Consumer Fraud and Trust Report today, and the survey of 1,029 United States adults places TikTok Shop ahead of eBay, Craigslist, Etsy and Poshmark as the platform respondents most often named when asked where a scam or counterfeit purchase happened.

In Short

A security company asked about a thousand American shoppers whether they had run into scams online, and a third of the ones who had named TikTok Shop as the place it happened. Most scams involved someone pretending to be a company the shopper trusted, and even though almost nobody blames the real brand, roughly three in ten say they will stop clicking that brand's ads afterwards. For advertisers, that means fraud committed by someone else can quietly shrink the audience willing to engage with their paid media.

A vendor survey with an advertising problem inside it

The report was published on September 9, 2026, from New York, and BrandShield describes it as an independent survey of 1,029 United States adults examining how scams, brand impersonation, counterfeit listings and AI-generated fraud are reshaping online shopping. The company sells digital risk protection services, which means the findings sit in the category of vendor-commissioned research: the firm publishing the numbers also sells the remedy the numbers imply. That does not make the data wrong. It does mean the framing deserves the same scrutiny an agency would apply to any supplier-funded study.

Methodological disclosure in the released material is thin. According to BrandShield, the sample is 1,029 United States adults. Fieldwork dates, margin of error, panel provider, question wording and weighting are not stated in the announcement. Several of the reported percentages also carry unclear denominators, a point examined below where it affects interpretation. The absence of those details limits how far any single figure can be pushed.

Concern about AI-generated fraud

Nearly 9 in 10 consumers, or 87.1%, say they are concerned that AI is making scams harder to detect, according to BrandShield, and 46.8% describe themselves as very concerned. On year-over-year sentiment, 52.3% say they are much more worried about online threats than a year ago and a further 29.8% are slightly more concerned, producing a combined 82.1%.

A discrepancy is worth naming. The pitch accompanying the report states that 9 out of 10 respondents claim AI has made scams harder to detect. The report itself measures something narrower: concern that AI is making scams harder to detect. Concern about a trend and a claim that the trend has occurred are different survey constructs, and the 87.1% figure supports the former reading only.

Yoav Keren, chief executive of BrandShield, framed the underlying mechanism in the announcement. "AI has made it cheap and fast to build convincing fakes, whether that's a phishing email, a customer support chat, a website, or an entire storefront, and the instincts people used to rely on to catch it don't hold up anymore," he said. "That puts real exposure on the brands at a time when customer trust and brand reputation are increasingly under attack."

The cost argument has independent support in prior reporting. Research from HUMAN Security documented that AI tooling has cut the entry cost of running a scam operation to roughly $5,000, with the components sold through open and dark web marketplaces. Where the fixed cost of building a convincing fake storefront falls, the volume of attempts rises, and detection systems built for a lower-volume era come under strain.

Exposure, losses and the base problem

More than half of respondents, 52.4%, say they were the target of a scam in the past 12 months. Within that population, 27.5% say they lost money outright. BrandShield reports that 4 in 10, or 43.3%, claim losses of $500 or more, though the announcement does not specify whether that share is calculated against all respondents, all targets, or only those who lost money. The three possible denominators produce materially different pictures of severity, and the released material does not resolve which applies.

Non-financial harms are reported alongside the losses: 23% had personal information stolen, 20% paid for an item that never arrived and 17.4% received a counterfeit. Those three outcomes map onto distinct fraud models. Data theft points to credential harvesting and phishing infrastructure. Non-delivery points to storefront fraud, where the transaction completes and the fulfilment never happens. Counterfeit receipt points to marketplace listing abuse, where a real product category is used as cover for an unauthorised good.

Independent government data gives the survey a reference frame. The Federal Trade Commission calculated that social media scams cost Americans $2.1 billion during 2025, with shopping scams the most frequently reported category and more than 40% of losses beginning with an order placed from a social media advertisement. Separately, the FBI's Internet Crime Complaint Center attributed $893 million in 2025 losses to AI-linked fraud. Both datasets are complaint-based rather than survey-based, so they undercount, but they establish that the phenomenon BrandShield's respondents describe is measurable outside a panel.

The impersonation finding that lands on media budgets

Among consumers who encountered suspicious activity or a scam, 63.6% say it involved someone impersonating a real company, bank, retailer or other organisation they trusted. The most commonly impersonated entities were companies or brands at 21.7%, retailers at 16.2% and banks at 14.5%.

What follows is the section of the report with the most direct relevance to paid media. According to BrandShield, 42.4% of consumers say they will now only buy from brand-affiliated websites, 39.2% will no longer share personal information with that brand online, 31.6% will stop clicking its ads, and 30.2% say they will stop buying from it online altogether. Against those numbers sits a single counterweight: just 4.2% of consumers hold the impersonated brand responsible for the scam.

That gap is the report's most useful structural observation. Blame and behaviour diverge. The brand is exonerated in attribution and penalised in conduct, which means the damage does not show up in any brand-health metric built around fault or sentiment toward the company. It shows up in click-through rate, in direct traffic share, and in the proportion of a retargeting pool that has quietly stopped engaging. A 31.6% stated withdrawal from clicking a brand's advertising, if it converts into behaviour at even a fraction of the stated rate, is a media efficiency problem attributable to an actor the advertiser never transacted with.

Two caveats apply. These are stated intentions collected in a survey, not observed behaviour, and stated intent consistently overstates action in consumer research. The figures also describe consumers who encountered impersonation of a specific brand, not the general population, so they cannot be read as a market-wide discount on advertising performance.

Keren returned to the same theme later in the announcement. "AI has made scams nearly impossible to spot, and our data shows consumers are willing to walk away from brands being impersonated even when they know it's not their fault," he said. "Protecting customers from fraud before it reaches them should be a core part of every brand's strategy to defend its reputation and maintain trust for long-term revenue growth."

The marketplace ranking, and what it does not control for

Among consumers who have used a marketplace platform, more than a third, 34.5%, say they have experienced fraud, a scam or a counterfeit item. The platform breakdown that follows is the finding most likely to travel: 34.7% ranked TikTok Shop as the platform where the fraud occurred, followed by eBay at 28.4%, Craigslist at 19.9%, Etsy at 19.2%, Poshmark at 17.0%, Mercari at 12.9% and Depop at 11.1%.

Those seven figures sum to 143.2%, which indicates a multiple-response question rather than a forced single choice. The announcement describes the base once as marketplace users and once as respondents surveyed, and the two are not the same population. Neither reading is confirmed in the released material.

More consequentially, the ranking is unadjusted for platform usage. A marketplace with a larger active shopper base will generate more fraud reports in absolute survey terms than a smaller one, even at an identical incident rate per transaction. Nothing in the announcement normalises the percentages against how many respondents actually shopped on each platform, which means the table measures where surveyed people encountered fraud rather than where fraud is most likely per purchase. Those are different claims, and only the first is supported.

Scale context matters here. TikTok Shop reached $15.82 billion in United States e-commerce sales during 2025, growing 108% year over year, and held an 18.2% share of the US social commerce market, according to EMARKETER, with the platform projected to reach $23.41 billion in 2026. Growth of that speed brings a large volume of new sellers, new buyers and first-time marketplace transactions into a single environment within a short window, which is a structural condition under which fraud reports accumulate. eBay and Craigslist, by contrast, have had decades to accumulate both users and incidents, and their appearance in the second and third positions is consistent with age rather than any recent change.

Platform enforcement already in motion

TikTok Shop has been layering enforcement mechanisms through 2026. A dispute policy dated June 25, 2026 requires sellers to issue full refunds on verified counterfeit items without the customer returning the goods, with counterfeit goods to be discarded by the buyer rather than shipped back. The platform's Assurance Health Rating system blocks new listings and campaigns once a seller's score falls below defined thresholds, with intellectual property violations covering counterfeit listings and missing brand authorisation as one of eight violation categories.

Federal product safety requirements were folded into the same guidance. From July 8, 2026, products needing a Children's Product Certificate or General Certificate of Conformity must have certificate data filed electronically before goods clear customs. A Prohibited Content for Platform Safety document published on August 4, 2026 bars fraudulent content, counterfeit promotion and impersonation of individuals or organisations in shoppable videos and livestreams. The pattern, across those four documents, is compliance infrastructure arriving after scale rather than ahead of it, which is the same sequence social commerce platforms have followed more generally.

The survey does not evaluate whether those measures work. It captures consumer perception during a period in which they were being introduced, which limits how much the ranking can say about current platform conditions as opposed to accumulated experience.

Where consumers assign blame

Asked who they would hold most responsible for a scam on social media or a marketplace, respondents put the platform at 26.2%, almost level with the scammer at 26.5%. Themselves came in at 16.8%, the marketplace seller at 10.3% and the impersonated brand at 4.2%. Those five figures total 84%, and the announcement does not account for the remaining share.

Behaviour changed for most marketplace shoppers who encountered a problem. Of the 79.5% who say a scam or attack changed how they shop, 28.6% now buy only from trusted sellers or official brand sites, 27.6% verify sellers before buying, and 10.6% stopped using the platform entirely. Just 20.4% report no change.

The near-parity between platform blame and scammer blame is the notable result. Consumers are not treating marketplaces as neutral venues in which a third party committed a crime. They are assigning roughly equal responsibility to the operator, which is the same logic regulators have applied in advertising.

The regulatory line converging on the same point

Ofcom opened a consultation on July 10, 2026 proposing Fraudulent Advertising Codes of Practice with penalties reaching 18 million pounds or 10% of global revenue, covering account bans, financial services verification, advertising moderation and the testing of AI advert-generation tools. The regulator listed impersonation of public figures and brands, cloaked landing pages and the use of generative AI among the risk characteristics providers should learn to recognise.

Litigation has moved on a parallel track. The Consumer Federation of America filed a class action against Meta on April 21, 2026, alleging the company profited from scam advertising while making user-facing representations about protection. Platform enforcement disclosures give a sense of volume: Google permanently suspended more than 700,000 advertiser accounts promoting AI-generated public figure impersonation and suspended 39 million advertiser accounts in total during 2024, while Microsoft Advertising removed or restricted over one billion advertisements across the same year, with policy revisions in October 2024 aimed specifically at deepfake technology.

The marketing industry has also been a direct target rather than only a bystander. A phishing operation using OpenAI's name solicited fake advertising beta testers through Apple's TestFlight platform in January 2026, and a separate campaign targeted Google Ads agencies with fabricated high-value client leads routed through look-alike domains that redirected to the real company's website. On the supply side, the SourTrade operation assembled malware inside victims' browsers across 12 countries while impersonating TradingView, Solana and Luno, running through Google, Meta and X advertising infrastructure.

What the numbers change for advertisers

The report's practical contribution is not the marketplace ranking, which is unadjusted and contested by construction. It is the quantification of a cost that has been discussed qualitatively for years: impersonation damage that lands on the impersonated party's media performance without ever appearing in its brand tracker.

Three specific consequences follow from the data as published. The withdrawal from clicking advertising, at 31.6%, describes attrition inside the addressable audience rather than a shift in sentiment. The move toward buying only from brand-affiliated websites, at 42.4%, describes a channel shift away from marketplaces and toward owned properties, which changes where a retail media budget produces returns. And the 4.2% blame figure explains why the first two have gone largely unmeasured: a brand asking consumers whether they hold it responsible will hear no, while the same consumers reduce their engagement.

For platform selection, the ranking's limitations cut both ways. It provides no per-transaction incident rate, so it cannot support a comparative risk judgement between marketplaces. What it does establish is that a substantial minority of marketplace shoppers, 34.5%, report a fraud, scam or counterfeit encounter at some point, and that 10.6% of affected shoppers stopped using the platform entirely. Platform abandonment at that rate has consequences for reach on any marketplace where a brand buys media.

The survey arrives as the underlying economics keep moving in the fraudsters' favour. Detection costs rise with volume, generation costs fall with model access, and the consumer response documented here is a withdrawal of engagement from parties who did nothing wrong. Whether the 31.6% figure converts into observable click behaviour is the question a single survey cannot settle, and BrandShield has not published the fieldwork detail that would allow anyone else to test it.

Timeline

Summary

Who. BrandShield, a digital risk protection firm led by chief executive Yoav Keren, surveying 1,029 United States adults. The findings concern consumers, marketplace operators including TikTok Shop, eBay, Craigslist, Etsy, Poshmark, Mercari and Depop, and the brands whose identities are used in impersonation scams.

What. The 2026 Consumer Fraud and Trust Report, reporting that 87.1% of respondents are concerned AI is making scams harder to detect, 52.4% were targeted by a scam in the past 12 months, 63.6% of those who encountered suspicious activity say it involved brand impersonation, and 34.7% named TikTok Shop as the marketplace where fraud occurred. Among consumers whose trusted brand was impersonated, 31.6% say they will stop clicking that brand's ads while only 4.2% hold the brand responsible.

When. Published September 9, 2026. Fieldwork dates are not disclosed in the announcement.

Where. United States, with respondents drawn from the adult population and questions covering social media platforms and online marketplaces.

Why. Falling generation costs for convincing fakes have increased the volume of impersonation attempts, and the survey quantifies a consequence that brand-health measurement does not capture: consumers withdraw engagement from impersonated brands without assigning them blame, producing advertising attrition attributable to a third party. The marketplace ranking is unadjusted for platform usage and cannot support a per-transaction risk comparison.