Skip to main content
Skip to content

Site Updates

Follow along as we add new data, tools, and improvements to the database. This site is actively maintained and updated regularly.

New

iMessage Viewer: 4,509 Forensic Text Messages

New 3-panel forensic iMessage viewer at /messages/epstein. Browse 4,509 messages across 15 conversation threads from Epstein's iPhone (Cellebrite extraction). Features virtualized scrolling for large threads, full-text search, and a real-time cross-reference engine showing same-day flights, schedule entries, and documents for any selected message.

4,509
Messages
15
Threads
Jan 2017 - Dec 2018
Period
  • 15 conversation threads: Bannon (3,356), Walker (532), Scaramucci (107), Lajcak (103), and 11 more
  • Cross-reference sidebar: click any message to see same-day flights, schedule entries, and documents
  • Full-text search with Fuse.js across individual threads or all conversations
  • Virtualized scrolling (TanStack Virtual) handles the 3,356-message Bannon thread smoothly
New

Photo Evidence Gallery: 18,308 DOJ Photos with Face Detection

New photo evidence browser at /photos showing 18,308 photos from DOJ EFTA releases. 473 persons identified via Amazon Rekognition facial recognition across 624 photos with 975 face bounding boxes. Filter by person, release batch, or AI-generated description.

18,308
Photos
473
Face IDs
975
Detections
  • 18,308 photos from all 12 DOJ EFTA datasets with AI-generated descriptions
  • 473 persons identified via Amazon Rekognition with confidence scoring
  • Filter by person, release batch (VOL00001-VOL00010), and description search
  • Face detection tags link directly to person profiles (validated against DB)
New

Email Archive: 405,693 Searchable Email Records

New email archive browser at /emails/archive with 405,693 email metadata records from Epstein's personal accounts. Full-text search on subjects, senders, and participants. Filter by account (Gmail, Yahoo), sender direction, date range, and promotional status. Server-side pagination with PostgreSQL tsvector/GIN indexes.

405,693
Emails
5
Accounts
1990-2024
Span
  • 405,693 email records spanning 1990-2024 from Gmail, Yahoo, and associate accounts
  • Full-text search powered by PostgreSQL tsvector/GIN indexes
  • Filter by account, Epstein sent/received, promotional, and date range
  • Detail modal shows full recipient lists, CC/BCC, folder path, and release batch
Data Added

17 New Persons: Deutsche Bank Team + Key Associates

Added 17 persons discovered through deep-dive investigation of 405K email archive and iMessage data. Includes the complete Deutsche Bank team that serviced Epstein's accounts (2013-2019), Epstein's personal trader, AmEx travel concierge, Leon Black's CIO, and more. 2,686 documents auto-linked.

17
Persons Added
2,686
Docs Linked
1,515
Total Persons
  • Chip Packard: Head of DB Wealth Management Americas, approved Epstein as client, named in NYDFS $150M fine
  • Tazia Smith: DB Managing Director, Epstein's Client Advisor on $91M+ portfolio
  • Paul Barrett: Epstein's personal trader, Alpha Group Capital, power of attorney over accounts
  • Natalia Molotkova: AmEx Centurion concierge who booked Epstein's travel (CBS News)
  • Eileen Alexanderson: CIO of Leon Black's family office, central to $158M relationship
  • 12 more Deutsche Bank personnel completing the full banking team picture
New

Daily Schedule: 13K Appointments Extracted

AI-extracted 13,092 structured appointment entries from Lesley Groff's scheduling emails spanning 859 dates (2004-2019). Browse Epstein's daily calendar with timeline and table views, filtered by activity type.

13,092
Schedule Entries
859
Unique Dates
185
Persons Linked
  • 13,092 schedule entries from 973 OCR documents via Claude Sonnet
  • 859 unique dates spanning August 2004 to July 2019
  • 8,273 entries linked to 185 identified persons
  • Activity types: appointments, meals, meetings, travel, massages, phone calls
  • Person cross-referencing: see who Epstein met and when
  • Top persons: Leon Black (466), Ehud Barak (364), Woody Allen (311), Martin Nowak (254)
Data Added

Flight Logs Expanded to 3,615 + Migrated to Database

Added 1,907 extended flight logs (1991-2019) from House Oversight November 2025 release, more than doubling coverage. All flights migrated from static JSON to Neon Postgres with passenger-person linking.

3,615
Total Flights
7,286
Passenger Links
1991-2019
Date Range
  • 3,615 total flights (up from 1,708) covering 1991-2019
  • 7,286 passenger-flight links across 167 unique persons
  • 11 new frequent flyer persons researched and added
  • Flights now query-able via SQL in Neon (hybrid in-memory + DB)
Data Added

iMessages, Photos & Email Archive from Jmail

Browse 4,509 forensic iMessages from Epstein's iPhone with a 3-panel investigation viewer — cross-referenced with same-day flights, schedule entries, and documents. Plus 18,308 DOJ photos with AI face detection (473 persons, 975 bounding boxes).

4,509
iMessages
18,308
Photos
473
Face IDs
  • iMessage Viewer: 15 conversation threads with forensic chat interface and evidence sidebar
  • Cross-Reference Engine: select any message to see same-day flights, schedule, and documents
  • Photo Evidence Gallery: 18,308 DOJ photos with face detection, person filtering, and lightbox
  • 473 persons identified via Amazon Rekognition with confidence scoring
  • TanStack Virtual scrolling for the 3,356-message Bannon thread
New

Recovered Text Browser, Government Officials Search, Knowledge Graph + 19 Forensic Narratives

Four major additions: browse 38,705 pages of text recovered from under government redactions, search 132 government officials in the Epstein files, curated knowledge graph with 606 entities and 2,302 typed relationships, plus 19 forensic finance narratives added to Research Hub.

38,705
Recovered Pages
132
Officials
606
KG Entities
2,302
KG Edges
  • Recovered Text: 38,705 pages of hidden text from under DOJ redactions — financial records, emails, FBI reports, flight logs. Victim statements excluded per site policy.
  • Government Officials: 132 officials from executive branch, House, Senate, and judiciary searched across 1.38M documents. 4 with DIRECT personal connections.
  • Knowledge Graph: 606 curated entities with 2,302 typed relationships (traveled_with, communicated_with, paid_by, victim_of, etc.) ready for network visualization upgrade.
  • 19 forensic finance narratives added to Research Hub — Jeepers Pipeline, Leon Black's Money, Bear Stearns, Shell Map, The Accountant, and more.
New

Codename Decoder, DOJ Audit Tracker, Research Hub

Three major new tools: decode 60+ pseudonyms from the case files, track 114K+ documents removed from justice.gov, and browse 157 forensic investigation reports with 13,799 EFTA citations.

63
Codenames
114,221
Audit Docs
157
Reports
13,799
EFTA Citations
  • Codename Decoder: 60+ pseudonyms mapped to real identities with evidence timelines — "The Invisible Man" = Prince Andrew, "Mr. Big" = Leon Black, "Snow White" = Disney character codenames used by Jes Staley
  • DOJ Document Audit: 90,232 documents confirmed removed from justice.gov tracked with priority scoring, category breakdowns, and full searchable table
  • Research Hub: 157 forensic reports from Epstein Research covering financial forensics, individual dossiers, evidence analysis, intelligence connections, and debunked conspiracy theories
  • 8 missing financial entities added (PLB LLC, JMWT, Prytanee, Mort Inc, JSC Interiors, Insurance Trust) — community-reported gap
Improved

Premium Investigation Reports & Mobile UX

Court-ready PDF/HTML export for investigation reports with premium styling. Enhanced mobile calls-to-action and donation flows.

  • Premium PDF exports with cover page, executive summary, financial profile, and cross-check results
  • HTML report export for embedding in external publications
  • Mobile CTA improvements: floating tip buttons, donation bubbles, responsive layouts
New

Panama Papers, Paradise Papers & Offshore Leaks Cross-Reference

Every person and financial entity in the database has been cross-referenced against the ICIJ Offshore Leaks Database — 814K+ shell companies, 771K+ officers, and 25K+ intermediaries from the Panama Papers, Paradise Papers, Pandora Papers, and Bahamas Leaks. 6,435 matches found, 114 high-confidence exact matches. Results appear on person pages as "Offshore Connections" with dataset badges, jurisdiction data, and direct links to ICIJ.

6,435
ICIJ Matches
114
Exact Matches
5
Datasets
  • 6,435 matches against 5.4M ICIJ records across 5 datasets
  • 114 exact name matches including Bernard Arnault, Brad Wechsler, Conrad Black, Carlos Slim, Credit Suisse
  • Paradise Papers: 4,128 matches — the largest source
  • Panama Papers: 976 matches | Offshore Leaks: 559 | Bahamas Leaks: 458 | Pandora Papers: 314
  • New "Offshore Connections" section on person pages with dataset-colored badges and ICIJ links
  • Data sourced from CapitolGraph cross-reference (same Neon project, shared infrastructure)
New

FinCEN Suspicious Activity Reports Cross-Reference

All Epstein-connected persons and financial entities cross-referenced against FinCEN Suspicious Activity Reports. 5,019 SAR matches found across 36 unique entities including Deutsche Bank ($1.08B flagged), JPMorgan Chase, Credit Suisse, Barclays, Bank of New York Mellon, and Citibank. Connects case files to federal financial crime reporting.

5,019
SAR Matches
36
Entities
300
Exact
  • 5,019 SAR matches across 36 Epstein-connected entities
  • Bank of New York Mellon: 1,576 SAR matches — largest hit
  • Credit Suisse: 794 | Deutsche Bank: 521 | Barclays: 374 | Citibank: 188
  • JPMorgan Chase: $1.08B transaction flagged in SARs
  • 300 exact name matches confirmed
  • Data sourced from CapitolGraph FinCEN Files integration
Improved

Sourcing Methodology: Professional-Grade Verification Framework

Complete rewrite of the /sourcing page into a comprehensive, 12-section methodology document with live database statistics, 5-tier verification framework, 9-step data processing pipeline, victim privacy documentation, and legal framework — designed for journalist, legal, and law enforcement scrutiny.

12
Sections
12
Source Archives
5
Verification Tiers
  • 12 source archives documented with provenance citations and document counts
  • 5-tier verification framework: Court Document → Official Record → Sworn Testimony → Investigative → Unverified
  • 9-step data processing pipeline with live metrics (2.01M OCR records, 2.67M vectors, 1.38M hashes)
  • Victim privacy section detailing 16,924 PII redactions across 3 phases
  • Legal framework (First Amendment, EFTA, CVRA) and editorial standards
  • Table of contents, JSON-LD schema, OpenGraph metadata, breadcrumb navigation
New

Investigator Profiles: Streaks, Ranks, Expertise & Endorsements

Full investigator profile system with activity heatmaps, contribution streaks, rank badges, expertise tags, endorsements from other users, impact scores, discovery badges, and review statistics. Profiles showcase each community member's investigative contributions.

6
Profile Sections
5
Rank Levels
  • Activity heatmap showing daily contribution patterns over the past year
  • Contribution streaks and rank badges (Rookie → Lead Investigator)
  • Expertise tags derived from investigation focus areas
  • Endorsement system for peer recognition
  • Impact score aggregating review quality, discoveries, and community engagement
  • Discovery badges for milestone achievements
New

Legal Cases Expanded: 12 Cases, $853M+ Tracker, Evidence Packets

Major upgrade to the /cases section with 12 fully documented legal proceedings, an interactive $853M+ settlements tracker, 2025-2026 development timelines, and a PDF evidence packet generator for downloading case-specific document bundles.

12
Cases Documented
$853M+
Total Accountability
5
Categories
  • 12 case pages: USA v. Epstein, USA v. Maxwell, Giuffre v. Maxwell, Deutsche Bank, JPMorgan, USVI, and 6 more
  • Interactive settlements tracker totaling $853M+ in financial accountability
  • 2025-2026 developments added to every active case page
  • Evidence packet PDF generator for case-specific document export
  • Category filtering: Criminal, Civil, Regulatory, Legislative, International
  • JSON-LD LegalCase schema markup on every case page
New

Victim PII Redaction: 16,924 Redactions Across 1,567 Documents

Comprehensive victim privacy protection system with automated PII detection and redaction across three phases. Includes a 34-paragraph legal declaration with 14 exhibits, a fillable attorney PDF form for redaction requests, and a confidential removal portal.

16,924
PII Redactions
1,567
Documents
3
Phases
  • Phase 1: 624 redactions across 100 FBI FD-302 interview memos
  • Phase 2: 13,495 redactions across 927 critical documents
  • Phase 3: 2,726 redactions across 539 SSN-classified documents (6,968 false positives excluded)
  • Patterns: SSN, DOB, phone numbers, addresses, victim names (80+ public figures excluded)
  • Legal declaration with 14 exhibits documenting the full audit trail
  • Fillable PDF form for attorney-coordinated redaction requests
New

Forensic Finance System: $5.3B Traced Through 220+ Entities

Integrated the full forensic financial workbook from Randall Scott (CC BY 4.0) — 481 verified wire transfers, 228 classified entities, shell company hierarchies, and multi-bank coverage analysis. The /money page has been completely redesigned with 6 investigation tabs, and the AI assistant can now trace money through shell company chains.

$5.3B+
Total Traced
224
Entities
393
Verified Wires
190+
Shell Companies
29
Person Links
  • Ingested 393 verified wire transfers ($973M) from master wire ledger with Bates document references
  • 224 financial entities with shell company classification, trust tier hierarchy (Tier 1-4), and confidence scoring
  • New Wire Ledger tab — searchable, sortable table of all verified wires with CSV export and Bates links
  • Shell Network tab — 4-tier cascade visualization (Holding → Distribution → Operating → Personal)
  • Bank Coverage tab — volume analysis across JPMorgan, Bank of America, and more
  • Payment Types tab — interactive donut chart showing WIRE/FEDWIRE/CHIPS/ACH/CHECK distribution
  • 3 new AI chat tools: search financial flows, trace money flow, entity financial profile
  • Deep Analysis mode (Opus 4.6) for complex financial investigations
  • Person pages now show financial flows when entity-linked (29 persons connected)
  • API v2 updated with new filters: direction, payment_type, tier, bank, bates_ref
Improved

Africa Trip Passengers, AI Chat Drawer, and Bio Improvements

Added Africa trip flight passengers, relocated the AI Chat to a slide-out drawer for better UX, and improved bio sentence splitting to handle edge cases with international names.

  • Added Africa trip passenger data to flight records
  • AI Chat relocated to a slide-out drawer — cleaner integration, less intrusive
  • Bio sentence splitting now uses Intl.Segmenter for better handling of abbreviations and names
  • Submit Tip elevated to primary floating action button
New

Investigative News Bureau: 15 Deep-Dive Articles Published

Launched original investigative reporting section with 15 evidence-backed articles covering previously unreported connections. Each article cites specific EFTA document IDs and draws from the 2.1M+ document database. Also added 4 new entries to the Contradictions tracker and enriched bios and dossiers for 8 key persons.

15
Articles Published
200+
Document Citations
4
New Contradictions
8
Persons Enriched
  • 15 investigative articles published with 200+ document citations
  • Mottola Christmas emails: 29 EFTA documents reveal post-conviction relationship
  • Mitchell exposé: the man with 9,600 documents nobody has heard of
  • Barak tapes: 3-hour audio recording analysis with 60+ meeting timeline
  • Mandelson email correspondence: UK libel-safe documentary analysis
  • Indyke financial architecture: $1.3B in suspicious activity reports
  • Wechsler/Apollo: all three founders entangled, not just Leon Black
  • Clinton depositions: first presidential testimony in 40+ years
  • NPR missing pages: 53 Trump-related documents gap analysis
Data Added

March 5 DOJ Release Ingested

Ingested 1,000+ newly released pages from the DOJ March 5, 2026 document release, including FBI 302 interview memos, the complete 2006 case file, and 20 previously withheld documents.

  • FBI 302 memos: Trump accuser interviews (Aug-Oct 2019)
  • Complete 2006 Epstein-Maxwell case file (582 pages)
  • 15 docs incorrectly deemed duplicative now restored
  • 5 previously privileged documents released
Improved

Review System Overhaul: Ranks, Investigations & Impact Pipeline

Major overhaul of the crowdsourced review system. Fixed accuracy scoring (replaced broken gold-standard-only accuracy with consensus agreement rate), fixed rank promotion (4 Experts, 42+ Senior Analysts now properly ranked), added investigation pipeline with structured case files, consensus-to-action pipeline that creates AI leads and notifies reviewers, 8 new investigative badges, document flagging for investigations, and OCR auto-merge when 2+ users agree on corrections.

256
Reviewers Updated
4
New Experts
42+
Senior Analysts
8
New Badges
  • Fixed: All users showing 50% accuracy — now shows real consensus agreement rates
  • Fixed: Everyone stuck at Analyst rank — 4 promoted to Expert, 42+ to Senior Analyst
  • New: /investigations page with structured case files, evidence pinning, and certification
  • New: Consensus triggers AI leads + notifies all reviewers who contributed
  • New: Impact notifications — your review contributed to consensus, person confirmed
  • New: 8 investigative badges (First Consensus, Sharp Eye, Financial Forensics, Cross-Referencing, etc.)
  • New: Flag for Investigation button in review desk
  • New: OCR corrections auto-merge when 2+ users submit similar fixes
  • Improved: Leaderboard names clickable to profiles, rank progress shown in feedback
  • Improved: My Reviews page now linked from 4 places (was only on dashboard)
New

Connection Lab: Multi-Entity Investigation Workspace

New 6-tab investigation tool for analyzing connections between 2-20 persons simultaneously. Replaces the limited 2-person co-occurrence search with a full Maltego-style investigation workspace featuring interactive heatmap matrix, D3 graph visualization, temporal analysis, evidence explorer, Louvain cluster detection with power rankings, and progressive entity discovery.

3,899
Lines of Code
10 files
Components
6
Tabs
20
Max Persons
  • Connection Matrix: Interactive heatmap grid showing pairwise co-occurrence counts across all selected persons with sorting, heat mode toggle, and ranked pair table
  • Graph Visualization: D3 force-directed graph with 3 layouts (force/radial/circular), drag interaction, zoom, edge weights, and node sizing by connection strength
  • Timeline Analysis: 3 views — density chart, person-by-year heatmap, and filterable event list showing when selected persons co-appear in dated documents
  • Evidence Explorer: Detailed shared documents and flights for any selected pair with CSV/JSON export and quick pair switching
  • Cluster Analysis: Louvain community detection, Brandes betweenness centrality, clustering coefficient, and composite Power Score rankings for the selected group
  • Entity Explorer: Maltego-style progressive discovery — expand any person to see their top 30 connections from the full database, filter in-group vs outside, one-click add to lab
  • 4 presets: Inner Circle, Flight Regulars, Financial Network, Legal Players — instant group loading
  • 2 new API endpoints: /api/connection-lab (multi-person matrix + timeline) and /api/connection-lab/connections (single-person expansion)
Data Added

Extended Flight Logs: 1991-2019

Added 1,907 new flights from pre-processed pilot logbooks, FAA FOIA releases, and ADS-B tracking data. The flight database now spans from 1991 through 2019, with 94 resolved person connections across 3,615 total flights.

1,907
New Flights
3,615
Total Flights
1991-2019
Date Range
94
Resolved Persons
  • Added 510 pre-1997 flights (1991-1996) never previously indexed
  • Merged passenger manifests from two independent GitHub research datasets
  • Resolved 94 unique persons to flight records using multi-source cross-referencing
  • Extended aircraft tracking to include Hawker HS-125 (N108JE, N125JE) and Cessna Citation (N401CB)
Data Added

Community Submissions Review + Two Investigative Articles

Processed 26 community submissions, added 15 new persons, updated 6 existing profiles, merged 2 duplicates, and published two original investigative articles.

26
Submissions Reviewed
15
Persons Added
6
Profiles Updated
2
Articles Published
  • Published: Who Is Jennifer Turner? Inside Ghislaine Maxwell's Seven-Year Financial Alias
  • Published: Same Bank, Same Spreadsheet: How Deutsche Bank Managed Epstein and Trump Side by Side
  • Added 15 new persons including Louella Rabuyo (Epstein housekeeper, 173 docs), Jennifer Turner (Maxwell alias), Nili Priell (Barak liaison), David Gergen, Peggy Siegel
  • Merged Jeffrey H. Sloman duplicates (764 doc links) and Kathryn Ruemmler duplicates (9,118 doc links)
  • Updated Darren Indyke (Parlatore Law Group), Rosemary Vrablic (Trump banking), Boris Nikolic (March 2019 contact)
New

Forensic Analysis Engine + Investigation Board

Major upgrade: 12-tab forensic finance suite with graph algorithms, Benford's Law, structuring detection, geospatial money map. New investigation board for drag-and-drop investigations. External cross-check integration (ICIJ, OFAC, SEC, CourtListener). Court-ready PDF report generator.

12
Money Page Tabs
4
Forensic APIs
6
Graph Algorithms
18
Jurisdictions Mapped
  • 12-tab forensic financial analysis suite on /money
  • Graph algorithm engine: PageRank, betweenness centrality, risk scoring, community detection
  • Benford's Law chi-squared anomaly detection on transaction amounts
  • Multi-hop financial path finder with BFS traversal
  • Unified forensic timeline merging flights, wires, emails, documents, legal events
  • Structuring detection: SAR threshold analysis ($3K/$5K/$10K)
  • Geospatial money map with animated flow arcs across 18 jurisdictions
  • Investigation Board: drag-and-drop canvas workspace at /investigate
  • External cross-check: ICIJ Offshore Leaks, OFAC Sanctions, SEC EDGAR, CourtListener
  • Court-ready PDF investigation reports with 7 sections
  • Money and Entities pages restored to navigation
New

Forensic Financial Intelligence on Entity Pages

Entity pages for organizations linked to Epstein financial flows now display forensic financial intelligence dashboards. Money and Entities pages temporarily under construction as we finalize major visualization upgrades.

  • Money page and Entities page temporarily hidden from nav — major visualization upgrades in progress
  • 10 organizations enriched with forensic financial data (JPMorgan, Deutsche Bank, Citibank, Barclays, Bank of America, Wells Fargo, Apollo, Southern Trust, Gratitude America, Financial Trust Co.)
  • Entity cards show financial volume badges with sort-by-volume option
  • Top flows table with bank, payment type, and Bates reference columns
New

Follow the Money: Financial Flow Visualization

New interactive financial flow analysis page tracking $4.5B+ across 50 entities — banks, shell companies, trusts, properties, and hedge funds. Three visualization modes: Sankey diagram, force-directed network, and transaction timeline. Full filtering by flow type, amount, date range, and verification status.

$4.5B+
Tracked
50
Entities
53
Flows
12
Shell Cos
  • Interactive Sankey diagram showing money flows between 50+ entities
  • Force-directed network graph with zoom, pan, and entity selection
  • Transaction timeline with chronological flow visualization
  • Filter by flow type, amount threshold, date range, and verified status
  • Entity detail panels with connected flows and document evidence
Data Added

Data Quality Audit & Person Profile Fixes

Comprehensive audit of person profiles, fixing hallucinated bios, removing non-person entries, and adding researched descriptions for 16 key individuals.

3
Bios Fixed
34
Non-Persons Removed
16
New Bios Written
6
Categories Fixed
  • Fixed 3 hallucinated person bios (incorrect identities)
  • Removed 34 non-person entries (generic titles, organizations, prison roles)
  • Researched and wrote verified bios for 16 persons including Deutsche Bank bankers, defense attorneys, and key associates
  • Fixed 6 category mismatches (e.g., US Senator miscategorized as associate)
  • Corrected Jack Goldberg → Jack Goldberger name
New

Three Investigative Articles, New Persons, and Contradictions

Published three original investigative articles based on EFTA deep dives. Added Jackie Kalin, Mark Kaminski as persons. Enriched Ramsey Elkholy and Donald Trump profiles. Added 5 new contradictions (c-081 through c-085).

3
New Articles
2
New Persons
5
New Contradictions
73
Documents Linked
  • Published 'The Evidence Nobody Is Talking About: Seven Trump-Epstein Threads the Media Missed' (2,001 words)
  • Published 'The Model and the Morgue: How the Man Who Handled Korshunova's Estate Became Epstein's Chief Procurer' (1,854 words)
  • Published 'The Facilitator: How Eva Dubin Ran the Logistics of Epstein's Network From Inside Mount Sinai' (1,515 words)
  • Added Jackie Kalin (p-1928) with 40 document links, Mark Kaminski (p-1929) with 33 document links
  • Enriched Ramsey Elkholy profile: 3,026-document deep dive, structured timeline, key findings
  • Added 5 contradictions: Trump plane denial (c-081), Trump transparency act (c-082), Trump 15-year contact (c-083), Elkholy wake claim (c-084), Elkholy selfless claim (c-085)
  • Investigated victim redaction submissions, flagged 7 documents for Regina Feoktistova privacy
New

Survivor Privacy Protection

Added comprehensive privacy safeguards to protect survivors mentioned in the Epstein files.

  • Redacted survivor names from public profiles
  • Added survivor privacy request page
  • Filtered redacted profiles from directory listings
  • Community feed posts with unredacted victim info automatically scrubbed
Fixed

Media Gallery Fixes

Fixed broken images across the media gallery — replaced deleted Wikimedia URLs and corrected 473 Archive.org links.

473
Archive URLs Fixed
6
Wikimedia URLs Fixed
  • Fixed 473 Archive.org media items pointing to HTML pages instead of images
  • Replaced 6 broken Wikimedia Commons image URLs
  • Reclassified 206 document collections incorrectly labeled as photos
New

Survivor Privacy Protection + GSC Fixes

Comprehensive survivor privacy system: redacted 29 survivor/victim names from all documents, OCR text, and search indexes. Added redaction request page for survivors to request information removal. Fixed 892 Google Search Console 404 errors, 128 server errors, and structured data issues.

29
Survivors Protected
1,205+
Documents Redacted
892
404s Fixed
22
New Redirects
  • Survivor names redacted from 791 document summaries, 373 titles, 1,205 OCR texts
  • New /redaction-request page for survivors to submit removal requests
  • Survivor profile pages return 410 Gone with privacy notice
  • 22 person alias redirects (dershowitz → alan-dershowitz, etc.)
  • 2,041 EFTA documents migrated to database with embeddings
  • Fixed Events structured data and added canonical tags
  • Admin login fix (Cloudflare WAF exclusion)
  • Person page DB calls parallelized to prevent 5xx timeouts
New

Public REST API v2 — 27 Endpoints, OpenAPI Spec, Bulk Export

Complete overhaul of the public API modeled after ProPublica and ICIJ. 27 endpoints covering persons, documents, flights, emails, locations, organizations, network analysis, search, and bulk data export. Every response includes citation metadata, rate limiting, and an OpenAPI 3.1 specification.

27
Endpoints
10
Resource Groups
CSV + JSON
Export Formats
  • 27 REST endpoints across 10 resource groups: persons, documents, flights, emails, locations, organizations, network, search, stats, export
  • Full OpenAPI 3.1 specification at /api/v2/openapi.json — machine-readable, enables client SDK generation
  • Rate limiting via Upstash Redis: 100 requests/hour anonymous, 1,000/hour with API key (X-API-Key header)
  • Bulk CSV/JSON export for persons, flights, locations, and organizations at /api/v2/export
  • Every response includes citation metadata (source, license, access timestamp) for academic and journalistic use
  • Network analysis endpoints: co-occurrence between persons, shortest path finding, full graph data for visualization
  • Document OCR text and SHA256 integrity verification available via dedicated sub-endpoints
  • Interactive API documentation at /api-docs with code examples in curl, Python, and JavaScript
New

Entities & Organizations: Major Upgrade

The Entities section has been completely rebuilt with a Neon Postgres backend, automated NER-based organization discovery, document auto-linking across 2.1M+ documents, and rich detail pages with timelines, financial data, and document evidence panels. 49 organizations are now tracked with thousands of document connections each.

49
Organizations
10K+
Doc-Org Links
6,167
NER Entities
3
New Components
  • Migrated from static seed files to Neon Postgres with full-text search
  • 10 new organizations auto-discovered from NER entity extraction
  • Document auto-linking: scanning 2.1M+ docs for organization mentions via FTS
  • Rich detail pages: timeline events, financial data, document evidence panels
  • New admin API: CRUD operations for organizations with cache invalidation
  • Legal defense system: /legal-requests policy page, email templates, demand tracking
New

Traffic Milestone: 58M+ Requests in First Week on Cloudflare

Epstein Exposed surpassed 58 million HTTP requests, 1.68 million page views, and 867,000 unique visits in just 7 days — from 199 countries. The site also handled 2 million AI crawler requests from Meta, OpenAI, Anthropic, Amazon, Microsoft, Perplexity, Google, and Apple. All while running on a solo-operated stack: Vercel Pro, Neon Postgres, Cloudflare, and DigitalOcean.

58.88M
Requests (7d)
1.68M
Page Views
867K
Visits
199
Countries
  • 58.88M requests in 7 days across 199 countries
  • 1.68M page views, 867K unique visits
  • 1.51 TB data transfer
  • 2M AI crawler requests (Meta, OpenAI, Anthropic, Google, etc.)
  • Peak: 3M requests/hour during viral Reddit post (10M+ views)
  • Zero downtime throughout — solo operator
Improved

Person Pages Now Update Instantly

Person profile pages now read fresh data directly from the database on every render, so bio corrections and category changes appear immediately without requiring a full site redeploy.

  • Added getPersonFieldsFresh() — uncached DB query for mutable person fields
  • Applied to both person pages and OG social preview images
  • DB updates to bios, categories, and descriptions now appear within 5 minutes
New

Report Inaccuracies + Review Guide + Alias Search + Review UX

Users can now report wrong photos, bad bios, broken links, OCR errors, and other inaccuracies directly from person and document pages. A comprehensive review guide helps new document reviewers understand rating categories. Person search now matches aliases, and the review desk has improved help and visibility.

7
Report Categories
7
Guide Sections
5
Review Improvements
  • New 'Report Issue' button on every person and document page — submit inaccuracy reports with categorized issues (wrong photo, bad OCR, broken link, etc.)
  • Reports go to the admin panel and trigger email notifications via Resend for fast triage
  • New /review/guide page with rating decision tree, concrete examples for each rating level, all 13 tag descriptions, keyboard shortcuts, and FAQ
  • Person search now matches aliases — type a known alias to find the person (alias hint shown in results)
  • Review rating selector has a '?' help toggle showing concrete examples for each category (Routine vs Junk confusion resolved)
  • 'Suggest New Person' feature now highlighted in a green box with icon — much more discoverable for reviewers
  • Review desk has 'People' and 'Locations' section dividers plus guide/shortcuts links at bottom
Improved

Mobile Network Graph + EFTA ID Search + Document Filter Fixes

Complete overhaul of the network graph for mobile devices, plus direct EFTA document ID search so you can find any document by its ID instantly. Fixed broken source and category filter buttons in the document browser.

+21%
Mobile Graph Area
+78%
Desktop Graph Area
11
Broken Labels Fixed
  • Network graph is now fully usable on mobile — compact header, touch-optimized interactions, bottom-sheet details panel, performance-tuned node count
  • Search by EFTA document ID (e.g. EFTA01249597) now returns results instantly via direct database lookup instead of failing through full-text search
  • Fixed DS9 and other missing source filter buttons on the documents page — 8 broken source labels and 3 broken category labels corrected
  • Desktop network graph area increased 78% by collapsing filters by default
  • Body scroll locked on network page for proper full-viewport experience on all devices including iOS Safari
  • Disconnected nodes automatically hidden in core density mode for cleaner graph visualization
New

Hybrid Search with 2.67M Vector Embeddings + Ask AI

Every document chunk in the database now has a vector embedding for semantic search. The new hybrid search combines keyword matching (BM25) with AI-powered semantic understanding for dramatically better results. Plus a new inline 'Ask AI' feature lets you ask questions about any document.

2.67M
Chunks Embedded
1,536
Embedding Dims
3
Search Modes
  • 2,669,382 document chunks embedded with OpenAI text-embedding-3-small (1536 dimensions) for semantic search across all 2.1M+ documents
  • Hybrid search pipeline: BM25 full-text + vector similarity + Cohere reranking — results show keyword (blue), semantic (purple), or both (green) match badges
  • New 'Ask AI' button on document pages — ask questions about any document and get cited answers powered by Claude
  • Technical artifacts page at /artifacts — browse extracted entities, redaction analysis, and OCR quality metrics
  • Interactive artifacts browser with filtering and search across document metadata
Improved

Enhanced Review System + Community Forum Self-Hosted

Major upgrade to the document review system with new tagging categories, AI-powered person suggestions, and location detection. The community forum has migrated to a self-hosted DigitalOcean VPS with new plugins and features.

  • Document review desk now includes person name suggestions from OCR text and automatic location spotting for geographic tagging
  • New review tags: financial records, witness testimony, travel records, and more granular categorization options
  • Community forum migrated to self-hosted DigitalOcean infrastructure for better performance and reliability
  • Forum now includes voting, gamification, calendar, data explorer, and reactions plugins — all free
  • Navigation revamped with improved document and person search in the command palette
  • Users can now delete their own comments on feed posts and dossiers
New

7 Investigative Articles + 3 New Persons + 10 Timeline Events

Major expansion of the /news section with seven new investigative articles covering previously unreported connections, plus new persons and timeline events documenting key moments in the case.

7
New Articles
3
New Persons
10
Timeline Events
  • Published investigation into Bruce Moskowitz — Epstein's personal physician, Trump health gatekeeper, and member of the Mar-a-Lago Trio with direct lines to Bill Gates
  • Published investigation into Dana Burns — dual assistant to both Ghislaine Maxwell and Jean-Luc Brunel, corporate officer of multiple Epstein entities, named in flight logs and police reports
  • 5 additional articles: Peter Mandelson arrest, Deepak Chopra 'bring your girls' emails, Brian Vickers NASCAR/Epstein ties, Congress six names analysis, FBI 'stand down' pattern (1996-2019)
  • 3 new persons added: Ike Perlmutter, Marc Sherman, Maritza Vasquez
  • 10 new timeline events spanning 2017-2026 including UN declaration, Maxwell page block motion, and Mandelson arrest
Fixed

Data Quality Cleanup: 249 Garbage Entries Removed

Comprehensive data quality sweep removed 249 invalid person entries across two rounds — MySpace usernames, generic court roles, single first names, and a fabricated 'Richard Epstein' record that had accumulated 26,410 false document links. 40 persons received new detailed descriptions backed by document evidence.

249
Entries Removed
26,410
False Links Removed
40
Descriptions Added
1,723
Clean Person Count
  • Removed 249 garbage entries: MySpace usernames (preston77, Chels-ifer), generic roles (Plaintiff, Warden, TOWN DRIVER), and single first names
  • Deleted fabricated 'Richard Epstein' entry that was conflating 3+ unrelated people across Deutsche Bank records, FINRA filings, and jail records — removed 26,410 false document links
  • 40 persons received new detailed descriptions generated from document evidence and cross-referenced sources
  • Person count cleaned to 1,723 verified identities
Fixed

Photo Corrections: 18 Wrong Person Images Fixed

Community audit by forum member Wja identified 18 person dossiers displaying photos of the wrong person — homonyms pulled from Wikimedia Commons. 7 photos replaced with correct images, 11 removed where no free-license photo of the correct person exists.

  • 7 photos replaced: Barron Hilton (was Conrad Hilton Sr.), Guy Lewis (was sports figure), Duke of Rutland, Viscount Cranborne, Viscount Astor, Alice Fisher (was nursing pioneer), Jack Lang (was Australian politician)
  • 11 wrong photos removed entirely — no correct free-license image available for Leon Black, Alberto Pinto, and others
  • Community-driven quality control: reported by forum user Wja via the discussion forum
Improved

PDF Download Button + SEO & Performance Fixes

Document pages now have a direct PDF download button. Plus fixes for SEO title duplication across 19 pages, proper OpenGraph tags on document pages for better social sharing, and performance improvements site-wide.

  • New 'Save PDF' download button on document detail pages for direct file download
  • Fixed title duplication on 19 pages that showed 'X | Epstein Exposed | Epstein Exposed'
  • Document pages now have proper OpenGraph tags for social media sharing previews
  • Site stats now update in real-time from the database instead of stale build-time counts
  • Person document counts now use accurate database queries (fix reported by forum user VigorousMilkyPenguin)
Data Added

90 New Persons from Community Submissions + 41K Document Links

The largest community data contribution yet: 90 new persons have been added to the database from verified community submissions, along with 41,000+ new document-person links generated via full-text search against 2M+ OCR documents.

90
New Persons
41,453
New Doc Links
1,633
Total Persons
431
Submissions Reviewed
  • 90 new persons added from community submissions — researchers, associates, academics, and public figures identified by the community
  • 41,453 new document-person links generated via OCR full-text search enrichment across 2M+ documents
  • Fixed missing person links on key EFTA documents including the Barak-Summers-Epstein recording (EFTA01621008)
  • Person count updated to 1,633 across all site pages, SEO metadata, and AI agent prompts
  • All new persons are searchable and have cross-referenced document connections
New

Entities & Organizations + 12 Verified Persons

New Entities section for tracking organizations, corporations, and foundations connected to the Epstein network. Plus 12 new verified persons from community research with detailed evidence profiles.

35
Organizations
48
Person-Org Roles
12
New Persons
  • New /entities page — browse 35 organizations connected to the Epstein case with role hierarchies and inter-org relationships
  • Organization profiles show directors, officers, and documented relationships to other entities
  • 12 new verified persons added from community submissions with detailed evidence and EFTA document citations
  • Site performance improvements: fixed 145-second full-table scan on person document queries
  • Fixed empty database responses getting cached by Vercel edge for hours

Found something wrong? Have a suggestion?

We rely on the community to help us improve this database. Let us know if you spot an error, have a tip, or want to contribute data.

Get updates: