In a post dated September 18, 2026, Meta rejected the findings of a report issued by the Instrat Foundation on fraudulent advertisements on its platforms in Poland, saying the report’s estimates of revenue from these advertisements were based on imprecise definitions and assumptions that do not represent the Polish market or the experience of the average user.
The dispute comes in the broader context of advertising platforms’ ability to prevent fraud before it reaches users, as well as companies’ transparency in measuring the scale of the problem and disclosing the results of their monitoring and enforcement systems.
Meta’s Objections to the Report’s Methodology
Meta said the report treated every advertisement removed by the company as fraudulent, even though removal policies cover other categories, such as restricted goods and services, intellectual-property violations, and advertising-quality or formatting violations. The company believes this conflation inflates estimates of fraudulent-advertising revenue from the outset.
It also objected to the method used to calculate reach and cost. According to Meta, the report used reach data covering the entire European Union to estimate the number of users in Poland and assumed that each advertisement was viewed three times, based on an estimate drawn from a sample of 2,800 customers of a U.S. company, without the data being specific to Poland or to fraudulent advertisements. As for the cost per thousand impressions, it relied on a figure of 20.65 zlotys, while the report’s authors acknowledged that they did not have a direct empirical estimate.
Meta says some of the conclusions were based on only 108 advertisements observed from a single iOS device and an account that had been set up in advance to attract specific types of advertisements by viewing and clicking on them. The company believes this sample does not represent the typical Polish user. It added that the report generalized the results of only three days, separated by months, even though the data differed by as much as 80% between those days, and did so without a margin of error or statistical adjustments when converting them into an annual estimate.
What Measures Has Meta Announced in Poland?
The company said that a new artificial-intelligence system to combat the impersonation of public figures had entered service in Poland, among the first group of European countries to receive it, and that it removed more than 50% more advertisements impersonating well-known figures than the previous system.
Meta also announced that it would require identity verification for all financial-services advertisers targeting Poland. This is linked to the company’s global goal of having 90% of its advertising revenue worldwide come from verified advertisers by the end of 2026.
The company also launched an anti-fraud center on Facebook and Instagram, featuring guidance on recognizing and reporting scams, as well as a Meta AI assistant that can analyze recent interactions to help users find the advertisement, account, or content associated with the incident.
The Figures Cited by the Company
Meta says it removed 137,000 fraudulent advertisements in Poland between July 2025 and June 2026, and that more than 88% of them were removed before any user reported them. It also said that the rate of reports of fraudulent advertisements per advertisement view fell by 83% between July 2024 and June 2026.
These figures represent data disclosed by Meta and do not include independent details in the material about how they were audited or the definition of a fraudulent advertisement used in the measurement. Therefore, the dispute concerns not only the number of advertisements removed, but also the distinction between advertising-policy violations and proven fraud, as well as the extent to which the samples and estimates represent the actual scale of the problem.