Poland: Thin Evidence

Table of Contents

TLDR

Poland established Investigation Team No. 5 to probe Epstein's trafficking and Russian intelligence connections, but the public corpus contains almost no legitimate Polish financial evidence. Of 6,876 "Poland" mentions, the vast majority are scanning hallucinations from blank Deutsche Bank forms. The only "Krakow" reference is a confirmed false positive. Poland's best investigative path runs through the Butterfly Trust's sealed beneficiary names, not through our corpus (PAPER TRAIL Project, 2026a; PAPER TRAIL Project, 2026b).

The Investigation

On February 24, 2026, Poland's National Prosecutor's Office established Investigation Team No. 5, targeting an organized criminal group of international nature active between 2005 and 2018. Three weeks earlier, Prime Minister Donald Tusk had announced an analytical task force chaired by Justice Minister Waldemar Zurek and Special Services Minister Tomasz Siemoniak (PAPER TRAIL Project, 2026a).

Poland's investigative angle is not financial. It is trafficking and intelligence. Tusk stated publicly that it is "increasingly likely" that Russian services co-organized elements of Epstein's operation (PAPER TRAIL Project, 2026a). This framing positions the Polish investigation at the intersection of human trafficking and geopolitical intelligence — a domain where our corpus of financial records, shipping invoices, and corporate filings has little to offer.

6,876 Ghosts

The corpus contains 6,876 entity mentions of "Poland" across 690 documents. At first glance, this looks like a substantial evidentiary base. It is not (PAPER TRAIL Project, 2026b).

OBS-5, one of our retracted observations, documented the source of the contamination. Deutsche Bank KYC (Know Your Customer) forms contain a pre-printed field label: "SSN (U.S. Persons/ Non-U.S. Persons)." When these forms are blank — never filled in, stamped "CLOSED" on December 17, 2019 — the automated text scanner misreads the label text. "Persons/Non" becomes "Poland." The hallucination is systematic: every blank KYC form with this field generates a phantom Polish reference (PAPER TRAIL Project, 2026b).

The result is that the vast majority of "Poland" mentions in the corpus are artifacts of automated text scanning failing on empty government forms. They tell us nothing about Poland. They tell us a great deal about the limitations of automated text extraction from bureaucratic documents.

Zero Krakow

The situation with "Krakow" is worse. OBS-6 documented a single cluster of Krakow references in the corpus, concentrated in a 190-page document: Ghislaine Maxwell's UBS account statements from February 2014 (Bates number SDNY_GM_00023306). The document header reads "Resource Management Account." Across 190 pages of repeated boilerplate, the scanner mangled this into "Resource Krakow" (PAPER TRAIL Project, 2026c).

Every "Krakow" reference in the corpus traces back to this single scanning error repeated across a single document's headers. There are zero legitimate mentions of Krakow in the Epstein files.

Both OBS-5 and OBS-6 were retracted and preserved as methodological lessons. They demonstrate two failure modes: hallucination on blank form labels and hallucination on repeated page headers. Both were caught by the verification pipeline before they could propagate into analytical conclusions (PAPER TRAIL Project, 2026b; PAPER TRAIL Project, 2026c).

What Is Real

Stripping away the scanning noise, the legitimate Polish references in the corpus are sparse (PAPER TRAIL Project, 2026d):

Flight log entries show Epstein's aircraft traveled to Poland on July 7, 2004 (100 entity mentions) and August 1, 2005 (42 mentions). These are real. They document that Epstein's aviation network included Polish destinations during the network's active period.

Nadia Marcinkova appears 149 times across 38 documents. She is Slovak-born, not Polish, though the names are sometimes conflated in coverage. Her presence in the corpus is well-documented through court filings and the Non-Prosecution Agreement, where she is named as a co-conspirator (United States v. Epstein, No. 08-80736, S.D. Fla. 2007).

Warsaw (391 mentions) and Gdansk (327 mentions) remain unaudited. The ratio of real references to scanning noise for these cities is unknown. Given the patterns observed with "Poland" and "Krakow," a high false-positive rate is likely.

The Butterfly Trust Path

Poland's most productive investigative avenue does not run through our corpus at all. It runs through the Butterfly Trust.

The NYDFS consent order documents that the Butterfly Trust disbursed $2.65 million through 120-plus wire transfers to women with Eastern European surnames for stated purposes including "tuition," "rent," "hotel expenses," and "immigration expenses." The beneficiary names are redacted in the public consent order (New York State Department of Financial Services, 2020).

If any of those 120-plus wire recipients were Polish nationals, it would constitute direct financial evidence linking Epstein's trust infrastructure to Polish victims or facilitators. Poland's prosecutors could request the unredacted beneficiary names from NYDFS through mutual legal assistance treaties. This path bypasses the corpus entirely and goes to the source.

An Honest Assessment

We are publishing this analysis because intellectual honesty requires it. When a country opens a criminal investigation and our corpus appears to contain nearly 7,000 relevant references, the responsible action is to disclose that those references are almost entirely noise (PAPER TRAIL Project, 2026b).

Poland's Investigation Team No. 5 is pursuing real leads — trafficking networks, intelligence connections, organizational structures active across 13 years. Our corpus is thin on all three. The flight log entries confirm Epstein visited Poland. The Butterfly Trust pattern suggests Eastern European financial connections exist in sealed records. Beyond that, we have scanning ghosts and blank forms.

The evidence Poland needs is not in the 2.1 million documents we processed. It is in the records that remain sealed, classified, or held by financial regulators who have not yet been compelled to share them.

References

New York State Department of Financial Services. (2020). Consent order: Deutsche Bank AG — Butterfly Trust disbursements [Regulatory filing]. https://www.dfs.ny.gov/reports_and_publications/press_releases/pr202007071

PAPER TRAIL Project. (2026a). Stakeholder analysis — STK-1: Poland, Investigation Team No. 5 [Data]. STAKEHOLDERS.md

PAPER TRAIL Project. (2026b). Observations — OBS-5: Poland scanning false positive from blank KYC forms [Data]. OBSERVATIONS.md

PAPER TRAIL Project. (2026c). Observations — OBS-6: Krakow scanning false positive from UBS headers [Data]. OBSERVATIONS.md

PAPER TRAIL Project. (2026d). Corpus search — Poland-related entity mentions [Data]. Database: epstein_files, entities table.

United States v. Epstein, No. 08-80736 (S.D. Fla. 2007).

U.S. Department of Justice. (2026). Epstein Records Library — Data Set 10, source KYC form EFTA01268833.pdf [Government release]. 01_DOJ_DataSets/DataSet_10/