EpsteinGPT: An Artificial Intelligence Search Platform For The Epstein Files

PostedAugust 5, 2026

UpdatedAugust 5, 2026

ByEpsteinWiki

EpsteinGPT is a commercial research platform that uses artificial intelligence to search and analyze documents from the public Epstein record. The service advertises conversational search, source citations, relationship mapping, timelines, image analysis, handwriting recognition, multilingual search, and research reports.

The project says it has archived approximately 341,000 files containing more than 3.5 million pages and 246 gigabytes of material. These figures and claims about complete processing are published by EpsteinGPT. Researchers should independently verify the scope of the collection before treating it as a complete inventory of every public Epstein record.

EpsteinGPT can help users locate possible evidence. It cannot establish that every generated answer is accurate. The original file, complete context, and document provenance remain essential.

Researchers can visit EpsteinGPT and compare its results with original evidence preserved by Epstein Data.

Snapshot

Name: EpsteinGPT

Website: EpsteinGPT

Research application: EpsteinGPT App

Founder identified by the site: Dave, a software engineer and DevOps architect

Status: Beta

Stated purpose: Artificial intelligence search and analysis of Epstein related government records

Stated archive size: Approximately 341,000 files, 3.5 million pages, and 246 gigabytes

Starting subscription price: $4.99 per month at the time of review

Important distinction: The site says it is not affiliated with OpenAI

What Is EpsteinGPT?

EpsteinGPT presents itself as a conversational research system for the Epstein files. Users can ask questions in ordinary language and receive generated answers that are intended to cite specific documents, pages, and media timestamps.

The platform says it uses several artificial intelligence systems, including Claude, Gemini, and open source tools. Its About page specifically states that the project is not affiliated with OpenAI despite using GPT in its name.

According to the site, the platform was created by a software engineer identified as Dave. The project describes itself as an independent transparency initiative pursuing nonprofit status. It also says there is no corporate sponsor, government funding, political organization, or venture capital behind the project.

These descriptions come from EpsteinGPT itself. The site should update its public organizational information if its legal status, funding, ownership, or fiscal sponsorship changes.

Conversational Document Research

The platform’s central feature is a chat interface that searches its indexed archive. A user can ask a broad question, request more detail, and continue with related questions.

EpsteinGPT says each response includes citations to the relevant source records. This is valuable because a generated answer should never stand alone. A usable research response needs to identify the exact file, page, passage, image, or timestamp that supports it.

A citation does not automatically make an answer correct. Researchers must confirm that the cited source actually supports the statement and that qualifying language has not been removed.

Search By Meaning

EpsteinGPT advertises semantic search, which ranks documents by meaning rather than exact words alone. This can help locate records that use different terms for the same person, place, event, or concept.

For example, a researcher searching for travel to the Caribbean may need results that mention the Virgin Islands, St. Thomas, or a specific airport rather than the word Caribbean.

Semantic search can broaden discovery, but it can also return loosely related material. Every result must be checked for relevance before it becomes part of a timeline or theory.

Knowledge Graphs And Relationship Mapping

The platform advertises a visual knowledge graph containing people, places, organizations, and events. Users can select an entity and explore other entities connected through indexed documents.

Knowledge graphs are useful for discovering repeated references. However, they can make weak associations look more meaningful than they are.

Two people may be linked because one mentioned the other in an email. They may appear in the same news clipping, address book, court filing, or evidence index. A line between two names does not prove a personal relationship, shared purpose, knowledge of abuse, or involvement in criminal conduct.

Researchers should identify the document that created each displayed connection. EpsteinWiki’s guide to mapping the Epstein network provides additional context for evaluating associations.

Timelines And Pattern Discovery

EpsteinGPT says users can generate timelines for people and topics. These timelines may combine flights, meetings, emails, legal events, and other dated records.

Chronological organization can expose overlaps that are difficult to see in separate documents. It can also create misleading certainty when dates are incomplete, estimated, or taken from unrelated sources.

Researchers should verify every timeline entry separately. They should record whether the date comes from an email header, filing date, flight log, calendar entry, witness statement, file metadata, or generated inference.

The platform also advertises pattern discovery that identifies people who appear together across documents. Coappearance is a research signal, not proof of a social group or coordinated activity.

Image, Video, Audio, And Handwriting Search

EpsteinGPT says it can analyze images, videos, audio recordings, handwritten notes, spreadsheets, emails, and PDFs. It also advertises image similarity search and descriptive visual search.

These tools could make media collections easier to examine. They also require careful validation.

Facial similarity systems can misidentify people. Handwriting recognition can produce incorrect names and numbers. Automated video descriptions can misunderstand scenes. Audio transcription can confuse speakers or omit uncertain words.

Any identification involving a person should be confirmed through additional evidence. Researchers should preserve the original media file and cite the relevant timestamp or image identifier.

Multilingual Search

The platform advertises search and automatic translation across more than 100 languages. This could help researchers locate information in foreign correspondence, legal records, and international documents.

Machine translation can alter legal meaning, names, idioms, and levels of certainty. A translated passage should be checked against the source language before publication, especially when it supports a serious allegation.

Claims About Archive Completeness

EpsteinGPT’s homepage says the system has read every document, watched every video, and analyzed every image from the Department of Justice release. The same page also says artificial intelligence processing is still in progress.

Those statements create an important qualification. Data acquisition may be complete according to the operator while full processing remains unfinished. Researchers should not assume that every file type has been analyzed to the same depth or that every possible result is available.

The site also says every file has been verified with a SHA 256 checksum against the original Department of Justice source. Checksums can confirm that two digital files are identical. They do not confirm that a government release is complete, correctly labeled, or free from redaction errors.

EpsteinGPT should be treated as one archive and search layer, not the final authority on the complete universe of Epstein records.

Pricing And Access

The EpsteinGPT pricing page lists several access levels.

The Explorer plan was listed at $4.99 per month. It included full search, access to twelve datasets, full document viewing, and ten artificial intelligence chats each day.

The Investigator plan was listed at $24.99 per month. It added unlimited artificial intelligence use, deeper context, priority processing, entity dossiers, team workspaces, and investigation reports.

Publisher access uses custom pricing and adds a commercial use license, public data access, and dedicated support.

The site also advertises ten free queries without signup and separate search credit packs. These prices were labeled introductory beta rates and may change.

The $497 Archive Download

EpsteinGPT separately sells what it calls the complete Epstein document library for a one time price of $497. The archive sales page describes a 246 gigabyte download containing approximately 341,000 files and more than 3.5 million pages.

The page says the files are consolidated, organized, deduplicated, and delivered in original formats. It also says the download address remains valid for 24 hours after purchase.

Potential buyers should compare this commercial package with free sources before purchasing. The original Department of Justice files are publicly available, and community archive projects provide additional mirrors and indexes.

Claims such as complete, forensic grade, and every public release should be independently tested against a documented inventory. Researchers should request information about file manifests, checksums, missing records, duplicate handling, version history, and refund rules before paying for a large archive.

Privacy And Research History

The EpsteinGPT Privacy Policy says the site may collect names, email addresses, company information, use case descriptions, IP addresses, browser information, referring sites, pages visited, and time spent.

The policy also says blog pages use Google advertising and that advertising partners may use cookies for personalized advertising. It states that information may be shared with service providers, advertising partners, or when legally required.

The platform advertises research memory, saved searches, pinned findings, conversation histories, team workspaces, and generated reports. However, the privacy policy reviewed was last updated on February 9, 2026 and still refers to waitlist data. It does not clearly explain in the reviewed text how research prompts, saved investigations, chat histories, uploaded images, or team workspace contents are retained and processed.

Investigative users should request clear answers before entering sensitive research. Do not upload sealed records, privileged material, confidential source information, private survivor data, or unlawfully obtained files.

Artificial Intelligence Answers Are Not Evidence

EpsteinGPT says its answers are sourced and self checking. These features can reduce errors, but they cannot eliminate them.

Artificial intelligence can select the wrong source, overlook contradictory records, combine separate people, misread optical character recognition text, or turn an allegation into an asserted fact.

Generated conclusions about guilt, intelligence connections, trafficking participation, or personal relationships require especially careful review. The source record must support the exact wording of the published claim.

Researchers can compare cited files with evidence pages on Epstein Data and the United States Department of Justice Epstein Library.

How To Verify An EpsteinGPT Answer

A reliable verification process should include several steps.

Save the complete question and generated answer.
Record every cited document identifier, page number, and timestamp.
Open the original source file.
Read the surrounding pages or complete message thread.
Confirm names, dates, quotations, locations, and document types manually.
Identify whether the passage contains a fact, allegation, opinion, hearsay statement, or investigative lead.
Look for contrary evidence and duplicate versions.
Compare the cited file with an independently preserved copy.
Link the final publication directly to the evidence.

Examples of evidence pages available through Epstein Data include EFTA00128780, EFTA01433023, EFTA01578684, EFTA01139414, and EFTA00151067.

How EpsteinGPT Supports EpsteinWiki Research

EpsteinGPT can help researchers develop search terms, locate possible evidence, build preliminary timelines, and identify recurring entities. It may be especially useful when a question spans several file formats.

These capabilities complement EpsteinWiki research on how Jeffrey Epstein’s business and trafficking system worked, Jeffrey Epstein’s financial infrastructure, and institutional failures in the Epstein case.

EpsteinGPT should guide researchers toward records. EpsteinWiki articles should cite the records themselves.

Key Takeaways

EpsteinGPT is a commercial artificial intelligence research platform for Epstein related records.
The platform says it has archived approximately 341,000 files containing more than 3.5 million pages.
Its features include conversational search, semantic search, knowledge graphs, timelines, image analysis, handwriting recognition, multilingual search, and reports.
The site says it is independent and is not affiliated with OpenAI.
EpsteinGPT identifies its founder only as Dave on its public About page.
The service is in beta and its published description says artificial intelligence processing is still underway.
Subscription access began at $4.99 per month during this review.
A separate 246 gigabyte archive download was offered for $497.
Claims about complete coverage and forensic grade organization require independent verification.
The reviewed privacy policy does not clearly address all research memory, prompt, upload, and workspace data described by the platform.
Knowledge graph links and repeated document appearances do not prove wrongdoing.
Every generated answer must be checked against the original evidence before publication.

Why EpsteinGPT Matters

Millions of pages cannot be meaningfully investigated through manual reading alone. Artificial intelligence search can help journalists and researchers find records, recognize patterns, and ask better questions.

The same technology can create authoritative sounding errors at enormous speed. EpsteinGPT is most valuable when it exposes its sources and users verify them.

The research tool may accelerate discovery. Only the evidence can support the conclusion.

Sources

Tags:

Investigation Tools