Jeffrey Epstein’s death on Aug. 10, 2019, sparked a flurry of conspiracy theories, and the release of Epstein’s purported suicide note on May 6, 2026, is a good bet to be fodder for more.

But Epstein’s death is only one facet of the convicted sex offender’s story to spawn and sustain conspiracy theories.

The Department of Justice has released more than 3 million publicly available documents related to the shadowy sex-trafficking networks surrounding Epstein. Journalists and researchers are working to make sense of the massive trove of data, but it is going slowly, and the interface built by the Department of Justice to the documents is unwieldy.

In response, some Americans have taken it upon themselves to dive into the archive. They are using artificial intelligence to develop platforms to make navigating the Epstein files easier and to conjure up new assessments of all the information.

As a scholar of online conspiratorial activity, I’m seeing that these tools are also helping conspiracy theorists craft their narratives.

Do-it-yourself conspiracy platforms

Because the Epstein files are a massive, unstructured dataset made up of PDF files, videos, photographs and other materials, these platforms make it easier for people to see connections where none exist.

Some of the platforms are intentionally masquerading as neutral, data-driven AI research tools but are actually designed by conspiracy theorists to encourage and amplify conspiracy thinking, leading to what I call “platform conspiracism.”

Epstein conspiracy theories often follow a classic logical fallacy known as “post hoc ergo propter hoc” — assuming that because event A happened before event B, event A must have caused event B. For example, in 2017, QAnon participants claimed that there was a secret cabal of satanic pedophiles trafficking children, so by this faulty logic, the subsequent Epstein revelations must be evidence that QAnon was right.

Some Epstein platform operators are supplementing their thinking with ideas from QAnon and other online conspiracy movements about cannibalism, satanism or the CIA’s experiments with mind control in the 1950s known as MK Ultra.

The platform conspiracists have a ready audience because many Americans are concerned about the vast tentacles of Epstein affiliates reaching into government, entertainment, academia and the tech industry. And of course many people simply want to know who is in the files and why. The unintended, or in some cases intended, consequences are that the do-it-yourself conspiracy platforms encourage paranoia and conspiracism.

Each time the Department of Justice releases or tries to not release a new crop of documents, it sparks widespread interest. Social media influencers, for example, immediately share videos of their own interpretations of the files.

Conspiracy masquerading as data analysis

One platform, called the WEBB, promises to use AI for “document intelligence” that can purportedly help researchers explore the Epstein files, flight logs, court documents and depositions.

Using a slick interface with literal red threads animating the screen as a reader moves their mouse, WEBB automates the messy data cleaning tasks that are required when working with unstructured data. The site says WEBB converts, optically records and indexes the files’ contents automatically, making the documents “structured, searchable intelligence.”

Named for Gary Webb, an investigative journalist who reported on an alleged CIA drug trafficking operation, WEBB invites users to imagine themselves as scrappy open-source intelligence researchers. It presents the platform as an objective resource that will be transparent about its functions. But even in the necessary first step of cleaning data, researchers make decisions that can steer the results.

One of the creators of WEBB is purported antisemitic conspiracy influencer Ian Carroll, who has appeared on Alex Jones’ Infowars and other far-right shows espousing conspiracy theories about Jewish cabals, 9/11 Trutherism, Epstein and Pizzagate, the discredited conspiracy theory in which child sex trafficking was allegedly happening at the Comet Ping Pong Pizzeria in Washington.

Carroll, or someone on his team, engages with people who mine the Epstein files through the WEBB interface, sharing their interpretations with his 1.4 million followers. Carroll also posts explainer videos showcasing his own conspiracy theory research with WEBB.

Other platforms such as Epstein Exposed and Epstein File Search offer similar platforms for “doing your own research” in the Epstein files. While they are less overtly conspiratorial, lacking endorsements from conspiracy theorists such as Candace Owens, such platforms use social media posts to encourage conspiracy research on their platforms.

An expanding web

WEBB also promises to begin adding datasets to its AI-powered platform that are related to other conspiracy theories, such as those about the 9/11 attacks and UFOs. Added “files” even include books of the Bible that were removed.

The WEBB team stated that their AI tool won’t hallucinate – create false or nonsensical content – because it is only trained on the file data, even though large language models routinely hallucinate because they are driven by probabilities.

It remains unclear how WEBB won’t hallucinate when it jumbles the Epstein files, JFK reports, 9/11 documents and Book of Enoch together. Carroll is now promoting a product called WEBB Enterprise, which presumably will include more access and tools for a fee.

Conspiracy of data

Whenever Americans are hungry for answers, platforms such as these can more easily masquerade as objective data analysis tools. They feed into a “conspiracy of data,” in which false or misleading information is presented in charts and graphics that create the impression of accuracy and authority.

Legitimate data analysis is complicated, messy and challenging, and best practices call for data analysis tools to emphasize transparency and context. For example, journalists at The New York Times use AI to supplement their work while also acknowledging the potential sloppiness of such tools and need for experts and journalists to do the work.

And as platforms like WEBB add their own datasets, the fodder for the paranoid fantasy of online conspiracy theories grows.

Matthew N. Hannah does not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.