Dataharvest 2026 - the European Investigative Journalism Conference

The opening of the conference will take place in the Aula Hanswijk (Z1.13, on the first floor), and will be streamed into the Aula Donche (Z1.15, first floor).

Drawing on real cross-border investigations from OCCRP's Research & Data team, this session will share the tools, techniques, and workflows the team relies on daily to support hundreds of journalists around the world. From geolocating images and tracking assets to social media investigations and smart browser hacks, this session will offer a practical, field-tested OSINT toolkit.

With shrinking newsroom budgets and a constant stream of "must-have" tools, it's harder than ever for journalists to know what OSINT tools might actually be worth using or paying for. This session cuts through the noise and focuses on what works right now.

Whether conference attendees are new to open-source research or looking for a sharp refresher, they will leave with concrete skills, trusted tools, and time-saving methods they can immediately apply to their own investigations.

This session will explain the methods behind an investigation into the European Defence Fund (EDF) that uncovered how structural weaknesses allowed Israel’s largest state-owned weapons manufacturer - directly involved in the war in Gaza- to receive millions in EU funding, despite rules meant to support only European companies. The reporters will explain how they identified relevant projects and traced the flow of funds.

They will break down our data work (scraping the tenders portal and building a dataset) and guide you through how the EDF policy works, its loopholes, legal framework, and how to work with the EU’s defence expenditure processes.

In today’s political environment of increasing militarisation, where a huge share of EU funding is directed toward defence, journalists need to learn how to access and analyse European public defence tender data, understand the EDF tendering and decision-making process, and see how European governments fund and benefit from defence projects.

This session will demonstrate how combining data work, investigative reporting, cross-border collaboration, and legal analysis can uncover hidden practices, reveal intentional gaps and inconsistencies that favour the arms industry over EU principles and international law.

This session will cover how to use trade information for investigative reporting, from the theory behind commercial flows to its application to real investigative cases. The first part of the presentation will focus on the "dictionary" that is crucial to read and interpret trade data. Then, it will explore how to source official customs statistics for free, understand crucial variables in import-export sheets, and find workarounds to expensive third-party commercial providers.

Examples from real investigations will show the power of using trade data in covering topics such as deforestation, sanctions evasion, the military industry, cocaine trafficking, but also much more "ordinary" commercial flows that might be linked to pollution/environmental issues. Throughout the session, participants will be welcome to bring examples of commodities they would like to track and guided in a few hands-on exercises to familiarise themselves with finding and understanding this kind of data.

The session is suitable for beginners. No technical/coding experience is needed, but the participants should be familiar with spreadsheets.

Most investigative workflows still rely on manually juggling dozens of tools. In this session, we'll walk through a live demo of a semi-automated pipeline built for real casework: web search and archiving with Playwright, face extraction, reverse image search, database cross-referencing with Telegram bots, social media analysis, and structured reporting via Obsidian mcp. All of this is orchestrated by Claude, an AI layer you can teach your own investigative methodology. At the end, participants will work through a simplified case using a workflow of their own.

Before the session, please install: Python, Claude Code. This session will teach participants to combine several smaller OSINT tools so they work together efficiently without requiring much manual effort. No special tools needed

Materials: https://github.com/anastasiiamorozova/ai-osint

"Agentic AI" is all the rage, but what does it offer beyond traditional LLM workflows? In this hands-on session we'll answer this question (and more) while leveraging Python's Pydantic AI library to build a start-to-finish agentic AI workflow.

Participants will learn how agents work, when they're useful, how to build custom tools, and options for tracing and evaluation. You'll leave able to write agentic workflows to extract information from texts, do semi-autonomous research, and deliver clean, structured results.

Basic experience with Python/LLMs is helpful but not required. After attending this session, participants will be able to understand when and how to apply agentic approaches to problems. Participants should have Python/Jupyter installed or a Google account for working in the cloud.

Having a GitHub account - https://github.com - will allow you to easily run the code without anything on your own computer.

Workshop materials can be found at https://jsoma.github.io/workshop-ai-agents/

Reporting online today, journalists have to battle astroturf campaigns, fake news sites and sketchy shell companies to find out who is behind the story. It frequently leads to a frustratingly common question: who is behind this website?

Popular tools and approaches to investigating websites have been less reliable lately. There's more opaqueness in areas where there should be more transparency; crypto payments add a layer of confusion, and generative AI makes it easy for adversarial actors to operate hundreds of websites.

Using a range of OSINT tools and real-world investigations, we will walk you through investigating the provenance and ownership of websites: identifying the scope and scale of the network it belongs to — if any? Who’s behind the site, now and in the past? Who are the main actors promoting this website? Is it AI slop? Are foreign actors likely behind the domain?

While it is not always possible to fully unmask the owner of a site, using a thorough checklist of tools and techniques that we have used in real-world investigations we can help you make sure to reveal as much as possible about a website, and potentially uncover important clues. We will also walk you through how to conduct these investigations safely depending on your threat model, and how to document your findings reliably.

 This session is suitable for beginners and doesn't assume existing technical knowledge.

In this session, we’ll show you how we approached a sensitive topic: mapping the potential exposure of all schools to the use of pesticides from surrounding agricultural activities. We will explain  how - working with scientists - we came up with a robust methodology,  found the data we needed and then crunched it to tell us where to go to make truly data-driven reporting on the ground.  

This investigation, published in Le Monde in December 2025, has sparked a lot of national and local interest thanks to its interactive map. Attendees will leave the session with a clear step-by-step guide to get started and adapt the ambition of the investigation to their own capacity. Feel free to bring with you any datasets that would help to map similar issues in your country.

You can read the main story in English here
Free access to the map in French here

You have a folder of documents and you want to extract data points from each one. And the data isn't in a structured table with neat rows and columns either. Here's where string functions and regular expressions can help. The demonstration will be in R but the skills are generic to all languages.

Materials: https://github.com/gebelo/Dataharvest2026

Say hello to Natural PDF, a new Python library for wrangling PDFs that's focused on usability and feature-completeness. Process PDFs with scraping-like selectors and spatially-aware queries, asking for "the red alphanumeric string" or "the content below the big Summary header." Beyond the basics, Natural PDF is also full of modern conveniences like table detection, multiple OCR engines, and citation-aware LLM data extraction.

To get the most out of this session, participants should have experience with Python and struggling with terrible PDFs.

Materials: https://jsoma.github.io/natural-pdf-workshop/