Meta disputes the methodology of the recent Instrat foundation report regarding fraudulent advertising on social platforms

The digital advertising landscape has become a primary battlefield for cybersecurity, pitting global technology giants against increasingly sophisticated criminal syndicates. This week, the Warsaw-based think tank Instrat published a report detailing the scale of fraudulent advertising on Meta’s platforms, sparking a sharp rebuttal from the tech conglomerate. The report, which purports to quantify the financial and social damage caused by scam advertisements, has been dismissed by Meta as a work of "fiction" designed to influence legal proceedings rather than provide an objective analysis of platform safety.

The controversy centers on whether the methods used to track, report, and remove malicious content are adequate, and whether current regulatory and voluntary measures are sufficient to protect the digital economy.

A Chronology of the Conflict

The tension between transparency advocates and Meta has been building for several years. In early 2024, a series of high-profile scams involving deepfake imagery of public figures began appearing across social media in Poland, leading to an outcry from both consumers and financial institutions. By mid-2024, the Polish Financial Supervision Authority (KNF) and local law enforcement agencies began coordinating with tech companies to improve reporting mechanisms.

The Instrat report, released this week, claims that the prevalence of these ads represents a systemic failure of Meta’s internal moderation systems. However, Meta points out that the report was commissioned by a law firm representing a plaintiff currently involved in litigation against the company. This contextual background, Meta argues, renders the report a tactical legal instrument rather than an independent scientific study.

Methodological Discrepancies and Data Integrity

At the heart of the disagreement is how the term "fraudulent advertisement" is defined and categorized. The Instrat report asserts that its methodology for estimating illicit revenue was based on identifying advertisements removed by Meta. Meta’s response argues that this premise is fundamentally flawed, as it conflates fraud with a broad spectrum of policy violations.

According to internal policy documents, Meta removes advertisements for a variety of reasons, including violations of intellectual property rights, non-compliance with regional financial regulations, formatting errors, and prohibited content categories. By treating every removal as a confirmed "fraudulent" instance, the researchers allegedly inflated the scale of the problem.

Furthermore, Meta has raised concerns regarding the statistical validity of the report’s projections. The study relied on a sample size of only 108 advertisements monitored on a single device running the iOS operating system. The account used for this observation was specifically curated—through repeated interactions and clicks—to attract high-risk content. Meta argues that this "honeypot" methodology fails to reflect the typical experience of an average Polish user, as the algorithms were effectively "trained" by the researchers to feed them the very scams they intended to measure.

The report also drew criticism for its temporal extrapolation. The findings regarding annual revenue loss were based on data collected over only three non-consecutive days in October, December, and January. Meta’s data science team noted that the volume of such ads varies significantly day-to-day, with a volatility rate of up to 80%. Relying on such limited snapshots to project annual trends without accounting for statistical margins of error is, according to industry standards, statistically unreliable.

The Scale of the Counter-Measures

While the report attempts to highlight a lack of oversight, Meta has provided counter-data to illustrate the scale of its proactive enforcement. Between July 2025 and June 2026, the company reported the removal of approximately 137,000 advertisements identified as fraudulent. Notably, 88% of these removals occurred automatically through machine learning systems before a single user report was filed.

The company further highlighted a significant downward trend in user-reported fraud. From July 2024 to June 2026, the rate of user reports regarding scams—calculated per impression—dropped by 83% in the Polish market. Meta officials suggest that this decline is evidence that their investments in AI-driven detection and increased cooperation with law enforcement agencies are yielding tangible results.

Broader Implications for the Digital Ecosystem

The standoff between the Instrat foundation and Meta underscores a deeper, systemic challenge facing the digital economy: the "verification gap." As criminals employ increasingly sophisticated AI tools to bypass traditional moderation, platforms are forced to strike a delicate balance between open advertising markets and stringent user protection.

Industry analysts suggest that the rise of "scam-as-a-service" models, where cybercriminals purchase infrastructure to deploy malicious campaigns, has outpaced traditional regulatory frameworks. The reliance on automated systems means that while the volume of removals is high, the "false positive" rate remains a point of contention for legitimate businesses whose ads may be caught in the sweep.

The implication for the average consumer is that digital literacy remains the final line of defense. Even as companies like Meta tighten their security protocols, the adaptability of modern fraud syndicates suggests that no single technological solution can fully eliminate the threat.

Expert Perspectives on Regulatory Oversight

Legal experts observing the situation note that the debate over the Instrat report reflects a growing trend of "litigation-based research." In this landscape, NGOs and legal firms collaborate to produce data that can be used to influence public policy or serve as evidence in civil lawsuits. While such reports serve a function in holding tech giants accountable, they also face intense scrutiny regarding their independence and objective neutrality.

"The challenge here is to ensure that valid concerns about safety aren’t eclipsed by biased data," says an analyst specializing in digital platform regulation. "We need a standard for reporting that all parties can agree upon—one that accounts for the complexity of global ad ecosystems without resorting to the kinds of algorithmic ‘poisoning’ that make the Instrat study so controversial."

Conclusion: The Path Forward

Meta has stated its commitment to continuing its current trajectory: investing in automated detection, strengthening partnerships with regional financial authorities, and increasing transparency through public reports. However, the company maintains that the narrative suggesting it is "willfully blind" to fraudulent activity is not only false but ignores the massive financial and human resources dedicated to the problem.

As the legal proceedings in Poland continue, the debate is likely to intensify. The incident serves as a reminder of the fragility of trust in the digital age. Whether the path forward lies in stricter government regulation, more robust corporate accountability, or a combination of both, one fact remains clear: the war against digital fraud is an ongoing evolution that requires constant vigilance, accurate data, and, perhaps most importantly, a shared understanding of what constitutes a "safe" digital environment.

Ultimately, the dispute highlights that the metrics of safety are just as important as the actions taken to ensure it. Without a standardized, transparent, and mutually agreed-upon method for measuring fraudulent activity, the public is left to navigate conflicting narratives, each claiming to represent the truth of the digital landscape. For now, the conflict serves as a case study in the complexities of governing the modern internet, where every data point is potentially a tool for policy advocacy, and every policy change is a response to a rapidly changing criminal threat.