Full Epstein Index: 1,000-10,000 Raw Investigative Text Records

A set of between 1,000 and 10,000 text records of raw witness statements and unverified allegations related to the Jeffrey Epstein case, compiled by theelderemo. Updated in late 2025, the data consists of OCR-processed documents sourced from FBI files and investigative journalism. The repository includes graphic material regarding sexual abuse and trafficking.

Use Cases

• Named Entity Recognition to extract names from raw witness statements
• Natural Language Processing to denoise and correct OCR-scanned legal text
• Topic modeling of investigative journalism and FBI document themes

Strengths

• 1,000 to 10,000 records
• MIT licensed for open redistribution
• Primary source witness statements

Limitations

• High OCR noise and character errors
• Contains unverified allegations
• Graphic content regarding sexual violence

Provenance

Source

FBI and investigative journalism documents

Collection Method

OCR scanning of physical or PDF documents

Freshness

Last updated December 2025.

Geography

United States