When the Justice Department blew past its December deadline to release unclassified materials from the Jeffrey Epstein prosecution, a Denmark-based data scientist named Tommy Carstensen decided to step in. He had barely paid attention to the case before. Now he maintains what may be the internet's most sophisticated archive of Epstein-related documents, spending up to 50 hours a week on the project alongside his full-time job.
Carstensen's initial indifference to Epstein shifted when he learned the government had failed to meet its legal obligation to declassify files by a court-set deadline. The realization prompted him to build an interactive archive that includes organized analyses of over one million documents, property maps, financial transaction graphics, and a facial recognition tool that allows anyone to search whether a face appears anywhere in the released files.
He is not working alone. A growing network of journalists, researchers, and volunteer data specialists have taken the Justice Department's raw document dumps and applied technical analysis to surface information that was never easily accessible. Their efforts have caught the attention of serious newsrooms and academic researchers.
The latest addition came earlier this month when Tristan Lee, founder of the non-profit Decoherence Media, published a searchable database of faces appearing in photographs within the Epstein files. The tool uses facial recognition technology to identify individuals who show up in images but are absent from Epstein's email records. Lee's team found more than 100 people not mentioned in the emails and nearly 200 others who have never been reported in connection with Epstein at all.
Lee acknowledged the legitimate concerns surrounding facial recognition technology, particularly its documented bias against non-white faces. To mitigate that problem, his team discarded any matches they were not confident in and applied a 99 percent similarity threshold borrowed from law enforcement standards. Decoherence cross-checked identifications using multiple recognition models and manual verification.
"Epstein's network is particularly well-suited for using facial recognition," Lee explained. "Many of the people in his network are notable in some way, prominent enough to have news articles written about them or be featured on company websites." The existence of nearly 20 years of Epstein's emails and communications also allowed researchers to cross-check photo matches against documented interactions.
One example emerged when the database uncovered additional photographs of Google co-founder Sergey Brin at Epstein's property on Guana Island. Email records show that Epstein associate Ghislaine Maxwell contacted Epstein in December 2006 to confirm Brin would be at the resort. Brin, still a board member at Alphabet, did not respond to requests for comment. Court documents suggest Brin had sought Epstein's advice on tax matters.
The motivation behind these archival efforts goes beyond simple documentation. Lee said he wanted to cut through the noise of conspiracy theories and viral misinformation that has surrounded Epstein's story online. "There's still so much confusion about who Jeffrey Epstein was, who was in his network, and what his crimes were," he noted. "We wanted to provide some clarity, to help regular people, as well as journalists and policymakers, better understand who is actually part of Epstein's social circle, and how these elite networks of power and influence actually operate."
Researchers have had to navigate significant complications in their archival work. The Justice Department previously removed or retroactively redacted documents, in some cases because it had failed to properly redact victim identities. Carstensen wrote code to monitor the Justice Department website for changes. His archive automatically redacts known victim names, family members of victims, and images of survivors and minors. He has been responsive to takedown requests from survivors.
Lee criticized the government's redaction practices as inadequate. "The DoJ did a disgraceful job with redactions," he said. "The worst example I saw was that they released nearly 100 naked photos of one outspoken victim." The Justice Department has acknowledged some errors in its release process but maintains it ultimately complied with the law requiring declassification. A departmental watchdog is now investigating.
Emma Best, co-founder of Distributed Denial of Secrets, oversees access to more than 20,000 unredacted emails from Epstein's Yahoo account. Best's organization has taken a more restrictive approach than most, limiting access to vetted journalists and researchers only. Those granted clearance are instructed to republish only emails relevant to their reporting and to assume everyone mentioned in the correspondence is a potential victim unless proven otherwise.
"We've kept the standard to access the Epstein emails higher than usual for many other datasets because it's especially sensitive and salacious," Best said. Those emails are being republished by Jmail, a browser-based archive developed by volunteer tech workers that combines materials from multiple sources.
Best expressed gratitude that most journalists have understood their responsibility to survivors. "We've been fortunate that most journalists understand the duty we have to Epstein's survivors," she said.
Author James Rodriguez: "These volunteers are performing the archival work the government should have done cleanly from the start, but at least they're doing it with more care for survivors than the Justice Department ever showed."
Comments