Investigating human rights abuses hidden in plain data
Whistleaks is a research initiative exploring how AI-powered tools can help journalists and human rights organisations detect evidence of abuses in leaked and breached datasets.
We're building open methodologies for a problem that's only growing — the gap between the volume of exposed data and the capacity to investigate it responsibly.
Join Our Research Network
We're assembling a small, committed group of partners to shape the future of ethical data investigation. If you work at the intersection of journalism, human rights, or technology — we'd like to hear from you.
Register Your InterestFeatured Investigations
Real-world cases demonstrating how leaked data, combined with ethical AI analysis, can drive accountability and protect human rights.
The Xinjiang Police Files: Inside the Mass Detention of Uyghurs
Thousands of photographs and documents taken from police servers gave the world its first unauthorised look inside Xinjiang's detention camps. Verifying them took fourteen newsrooms months.
Read Full Case Study
The Uber Files: 124,000 Records on Lobbying, Police Raids and Drivers
A former Uber lobbyist handed journalists the company's emails and messages from its most aggressive years. It took around 180 journalists in 29 countries to work through them.
Read Full Case Study
Suisse Secrets: Credit Suisse Accounts for Criminals and Human Rights Abusers
The largest leak ever from a major Swiss bank showed how a global lender held money for people accused of torture, trafficking and corruption. Journalists in Switzerland could not touch it.
Read Full Case StudyThe Challenge
Data breaches increasingly expose information relevant to human rights investigations — communications, financial flows, surveillance records. But investigative organisations lack the tools and resources to process this volume of data systematically.
26B+
Records exposed in data breaches in 2023 alone
<1%
Of leaked datasets systematically reviewed for human rights evidence
Weeks
Average time for a small team to manually triage a single major breach
The result is an investigative capacity gap: potential evidence of corruption, surveillance, and abuse goes unexamined simply because no one has the bandwidth to look.
Our Approach
We're building tools and methodologies at the intersection of AI, investigative journalism, and human rights research. Our work is guided by four principles.
Automated Triage
We're exploring NLP and pattern-recognition tools that can surface potentially significant records from large, unstructured datasets — reducing weeks of manual review to hours.
Ethical Framework
Every methodology we develop is guided by strict ethical protocols for handling sensitive data, protecting sources, and minimising harm to individuals named in breaches.
Human-in-the-Loop
AI assists but never replaces editorial judgment. Our tools are designed to flag and prioritise — the investigative decisions always remain with journalists and researchers.
Security First
We're building with end-to-end security in mind: air-gapped analysis environments, encrypted workflows, and minimal data retention as core architectural principles.
For Partners
Whistleaks is early-stage and collaborative by design. We're seeking partners who share our belief that the investigative capacity gap is a solvable problem.
Investigative Journalists
We're looking to collaborate with newsrooms and independent journalists working on stories involving leaked or breached data. We can offer tooling support and methodological guidance.
Human Rights Organisations
If your organisation handles large datasets as part of human rights documentation or accountability work, we're exploring how our tools might support your processes.
Researchers & Technologists
We welcome collaborations with researchers in NLP, information security, and digital rights who share our commitment to ethical, impact-driven work.
From the Blog
What we're learning as we build.
Register Your Interest
Whether you're a journalist, researcher, or human rights professional, tell us about yourself and how you'd like to collaborate.