DOJ Releases 3 Million Pages of Epstein Documents, Triggering Wave of Misinformation

The US Department of Justice’s January 30 release of more than 3 million pages of documents related to Jeffrey Epstein has unleashed a torrent of both legitimate investigation and dangerous misinformation across social media platforms.

The massive document dump includes correspondence, flight logs, and investigative materials connected to the convicted sex offender’s network of associates. While journalists and researchers have begun the painstaking work of analyzing these materials, the release has simultaneously triggered an explosion of misleading, manipulated, and AI-generated content.

DW Fact Check investigated several of the most viral false claims circulating online, revealing how sophisticated AI tools are being deployed to create convincing but entirely fabricated narratives about the case.

Among the most widespread falsehoods is the claim that Epstein is still alive and living in Israel. Posts purporting to show recent photos of a bearded Epstein walking down streets in Israel have garnered millions of views across platforms. One post alone reached over 5 million views on X (formerly Twitter).

However, these images show clear hallmarks of AI generation. A reverse image search reveals a larger, uncropped version displaying nonsensical street signs with gibberish text that doesn’t translate to legitimate Hebrew or Arabic. Other AI indicators include motion blur effects where the central figure appears sharp while surrounding people are blurry, and cloned details in supposed “bodyguards” with nearly identical features. Most tellingly, the larger version includes a Gemini watermark in the corner, and the image contains logical errors like traffic lights with green at the top instead of red.

The official record clearly shows Epstein died by suicide in his cell in August 2019 while awaiting trial, as confirmed by New York City’s chief medical examiner following an autopsy.

Another widespread fabrication involves a video supposedly showing Donald Trump and Jeffrey Epstein at a party surrounded by minors. One post sharing this clip garnered more than 1.2 million views on X, with similar versions proliferating across Instagram and TikTok.

DW’s investigation determined this video was created by animating a genuine 1997 photograph from a Victoria’s Secret fashion event in New York, where Trump and Epstein appeared with model Ingrid Seynhaeve. The original photo contained no children whatsoever. The viral clip shows clear indicators of AI manipulation, including unnaturally smooth skin textures, inconsistent facial expressions, and physically impossible body movements.

A particularly disturbing trend involves social media users attempting to “unblur” faces of alleged victims in redacted photos. Under the Epstein Files Transparency Act, the Department of Justice is required to redact information that could identify victims to protect their privacy and prevent retraumatization.

One post showing a user asking X’s AI chatbot Grok to “unblur the face of a girl who’s with Epstein” received over 17.6 million views, with numerous similar requests gaining thousands of views. These AI tools don’t actually reveal genuine identities but instead hallucinate and generate entirely fabricated faces, creating false leads and potentially further victimizing survivors.

“One of the problems with all of the AI-generated content and fake videos circulating among the real videos is it becomes very difficult to distinguish what is real,” said Courtney Radsch of the Center for Journalism and Liberty in an interview with DW.

A recent report by NewsGuard, which analyzes the credibility of online information, demonstrated how quickly AI tools can create convincing deepfakes implicating public figures in the Epstein case. When testing three AI systems, only ChatGPT refused to create such fakes, while Gemini hesitated and Grok generated them within seconds.

Disinformation expert Tommaso Canetta from the European Digital Media Observatory warned that the stakes are high: “We have people believing in things that are not real, but they look exactly like reality.”

As journalists continue the legitimate work of analyzing the actual Epstein documents, the parallel flood of AI-generated content threatens to undermine public understanding of a case that already sits at the intersection of wealth, power, and exploitation.