Posted 4/10/2022
Distributed Denial of Secrets (DDoSecrets) is a transparency collective. We’ve seen some questions and confusion recently as to what that means, so I’d like to elaborate. See also our about page, our Wikipedia page, and articles here and here.
When whistleblowers within an organization, or hackers that have gained access to files, want to get the word out, DDoSecrets can:
Announce the release to a wide audience
Check the files for personally identifying information
Package a release for public consumption, via torrents or web-search interfaces
Contact journalists at a wide range of publications
Collaborate with academic researchers
Share more sensitive documents with journalists and researchers, without making all data public
That’s our primary role: connectors and proliferators of data.
For most releases, a source contacts us anonymously. They describe what data they have, we have a conversation to establish veracity and the scope of documents, and negotiate a safe way to move the data to us. We typically do not know who our sources are, and do not have contact with them after they submit documents. Occasionally a source will ask to be credited publicly, in which case we verify their identity and attribute the leak to them.
We also mirror releases published by other groups. For example, ransomware groups sometimes publish their victims’ data to criminal forums when their (typically corporate) victims refuse to pay. Other criminals can then peruse the release looking for usernames and passwords, personal information, and anything else they can profit off of. By making a copy of these releases for journalists, we hope that some social good can come out of what would otherwise only benefit other bad actors.
We’ve also published datasets on behalf of other public organizations, notably including the .Win Network archive originally scraped by the Social Media Analysis Toolkit (SMAT) team, and the Patriot Front files, sourced by Unicorn Riot.
When considering a new release, we have four main outcomes: discard the data, publish it publicly in full, publish it only to journalists and researchers, or publish a redacted subset of the documents publicly.
When we publish a release publicly, we typically release the documents via torrent, with private backups to reseed the torrents if needed. When possible, we also release documents through websites like DDoS Search, which can make data easier to browse, explore, and analyze.
When we release data with limited distribution, journalists and researchers contact us (or vice versa), and we negotiate a way to share files with them. If it’s feasible to filter out personal information, then we’ll publish the subset of low-risk documents publicly. However, given our limited staff, and the large size of many of these datasets (millions of emails and documents), we frequently mark datasets for limited distribution to vetted reserchers.
In either case, we add a detailed summary of the release to our website and usually announce publication via our newsletter, our Telegram channel, Twitter, and whatever other platforms are useful.
Usually our role ends as a publisher, but when resources permit, we also work with some journalists and academics to analyze datasets. Sometimes this aid is short-term; when our releases include content like database dumps that not all newsrooms are equipped to investigate, we’ve been able to help some news groups run a SQL query to find relevant information and export it in a simpler format like a spreadsheet.
In other cases, our collaborations have been more involved. DDoSecrets has worked on months-long investigations with:
The Organized Crime and Corruption Reporting Project (OCCRP) on 29Leaks
The University of Southern Maine and Harvard’s Institute of Quantitative Social Science, resulting in this paper so far
65 journalists and 20 outlets, including Forbidden Stories, on Mining Secrets
Our role in these collaborations has ranged from coordinating who’s investigating which sections of a release and collating results, to providing infrastructure (wikis for coordination, interactive sites for document search, and virtual machines for analyzing datasets), to writing our own document parsers and aiding in network analysis.
Finally, DDoSecrets aggregates investigations into our releases, summarizing and linking to articles on our newsletter and our wiki entries. Some of us interview with journalists, speak on podcasts or on television, and present at conferences about our work, and what our releases and investigations have revealed.
If you’re a fan of what we do, please consider donating, volunteering, and helping spread information about our work. We’re a small team with limited funding, and every bit helps!