Dataharvest 2026 - the European Investigative Journalism Conference

Processing natural language is seen as the task that artificial intelligence is most adept at. However, as journalists and researchers, we need our technologies to be explainable, understandable, and deterministic. Because of this, not all artificial intelligence algorithms are well-suited for our work. And, when every company promises that their AI software is extraordinary, it's difficult to distinguish the empty promises from what the technology can actually do. Working on OpenAleph, an open-source tool for investigative journalism, has taught us a lot about processing natural language. We extract names of people and companies from raw text. We try to infer the language a text is written in. The names of places, cities, and countries are crucial to us, in order to situate data geographically. All of this is heavily reliant on algorithms. But not all algorithms are as good as getting us what we want!

In this session, we'll show you what works and what doesn't. Everything we demonstrate can be used independently of OpenAleph, and integrated into your own workflows. Some machine learning algorithms are excellent at getting us more insights from our data. In addition to this, data that we already have, or public data, can be harnessed to help us identify names of people and places, just based on similarity - no AI required!

Finally, we'll discuss how these approaches compare to using large language models and generative AI. This session is half teaching and discussing common solutions, half workshop. For the workshop part, bring a laptop running Python if possible.

In this hands-on workshop, we will unpack our year-long collaborative investigation into the US data-analytics company Palantir Technologies in Switzerland. We uncovered how the tech giant tried over multiple years to sell its products to various Swiss government institutions and how it failed nearly a dozen times.

At first, our written requests to the authorities all received the same answer: "We have no contracts with Palantir." We went on to extensively FOIA the authorities and found traces of just how hard the company had tried to wriggle its way into Swiss administration, including the Armed Forces. Documents from 59 FOIA requests across 41 federal offices, combined with additional research, exposed the corporate sales playbook of one of the most controversial companies in the world.

The workshop is built around three things we want to share: First, the specifics of the methodology: how we ran a cascading FOIA strategy keyed to the name "Palantir" rather than to specific documents, letting one office's release point us to the next. And how that approach finally surfaced our crucial piece of proof: a classified Swiss army assessment warning against the procurement of Palantir's products on data-sovereignty and reputational grounds. Second, what a year of reading Palantir's correspondence with authorities taught us about how the company actually sells: high-stakes meetings between Palantir executives and senior officials at international security forums and the World Economic Forum, pro-bono pitches that arrive in moments of public crisis, and the persistence with which the company cycles between federal offices after each rejection. Third - in hindsight, which aspects of our research and reporting proved most important once Palantir sued Republik before the Zurich Commercial Court. The interview in Zurich, the extensive right-of-reply process before publication, and the documentation of every exchange. We will share what the litigation has looked like from the inside, and what we would do differently in hindsight.

Links to the investigation:
• Part 1: https://www.republik.ch/2026/02/18/how-tenaciously-palantir-courted-switzerland
• Part 2: https://www.republik.ch/2025/12/09/warum-palantir-zum-risiko-fuer-die-schweiz-wird
• English adaptation on swissinfo.ch: https://www.swissinfo.ch/eng/war-peace/why-palantir-is-becoming-a-risky-bet-for-switzerland/9066633518:25

AI-generated scam advertisements have flooded European social media platforms, using deepfake videos, cloned voices, and fabricated news stories to lure thousands of victims into fraudulent investment schemes. While the EU's Digital Services Act (DSA) was designed to curb illegal and harmful content, these scams continue to spread at scale, exposing major gaps in platform enforcement and regulatory oversight.

This session will show how to investigate AI-driven scam ads and platform failure using publicly available tools, leaked material, and EU tech legislation. We will walk through how to find and identify and analyse AI-generated scam ads and deepfake content, use Meta's Ad Library to map scam campaigns, detect duplication and evasion tactics, and estimate scale. The session will also address how to collaborate effectively with civil society organisations and trusted flaggers to monitor platforms and access specialised expertise.

As part of the presentation, we'll also talk about how to investigate the Digital Services Act in practice, including transparency obligations, systemic-risk provisions and reporting mechanisms; how to document and report on failures of enforcement by platforms and public authorities, and how to connect platform-level analysis to human stories, including victims' experiences. That also includes how to responsibly handle leaked messages, call scripts, and fake trading platforms, balancing verification, security and ethical considerations.

The session will include concrete examples and live walkthroughs, showing how journalists can combine platform data, legal frameworks and human sources to hold tech companies and regulators accountable.