Digital Digging

Imagine walking past a fancy restaurant. The menu is posted in a glass case outside. Truffle pasta. Wagyu beef. You can see everything, read every description—but you can’t eat any of it. You’re just looking through glass.
That’s what most public data feels like online.
You find a website with exactly what you need. A map of power outages. A table of housing prices. It’s right there on the screen. But there’s no download button. No export. You can look, but you can’t take it home.
A 2022 study in Scientific Data found that over 60% of government datasets are “semi-open”—technically public, but locked behind technical barriers that only coders can break through. The data sits in an unlocked room, but the doorknob is a puzzle only an engineer can solve.
Until now.
Read also the story below



Get full access to Digital Digging with Henk van Ess at www.digitaldigging.org/subscribe

NEW PODCAST Henk here! It’s been a while. Today I’m sharing something unprecedented. The first documented case of a fully autonomous AI cyberattack.
A Chinese state-sponsored group turned Claude into an autonomous hacker: it executed 80 to 90 percent of their espionage campaign. Across 30 organizations.
No humans typing commands! Just an AI working at physically impossible speeds. Yup. Stealing data! All on its own!
The hack? Simple role-play. They told the AI they were legitimate security researchers. It believed them!


Get full access to Digital Digging with Henk van Ess at www.digitaldigging.org/subscribe

Henk van Ess's article DeepSeek: The Real Price of China's AI investigates DeepSeek, a Chinese AI company, whose seemingly inexpensive ChatGPT competitor masks a sophisticated system of state control. DeepSeek's AI, while appearing cost-effective on the surface, is demonstrably limited in independent thought and strictly adheres to Chinese Communist Party propaganda. Today’s episode guides are, in style, the virtual hosts of NotebookLM.



Get full access to Digital Digging with Henk van Ess at www.digitaldigging.org/subscribe

Ever wish a chatbot could actually see what’s in a PDF instead of just skimming the text? Well, Claude—the AI chatbot—is now able to analyse the visual information.
Imagine the potential. From financial analysts spotting hidden profit drops in charts, to journalists exposing fake health diagnoses a bit quicker with the help of chatbot-aided discoveries. Suddenly, AI isn’t just reading documents; it’s seeing the whole picture (literally).
In this episode, I’ll break down why this shift is a deal for researchers, journalists, and anyone who’s ever spent hours dissecting official documents to find the story in the data. Well, it’s not me speaking, but my two AI-assistants.
(Below you see what the problem is with analyzing PDF’s with chatbots - but not anymore.)

Don’t forget to read the full article about VisualPDF:



Get full access to Digital Digging with Henk van Ess at www.digitaldigging.org/subscribe

Paul Myers (BBC Verify) and me, Henk van Ess sat down in London to discuss a pressing issue: the shitification of Google. Yes, that’s right—Google has slowly been losing its touch. But hold on—Paul wants me to clarify that these are his own spicy hot takes, and in no way represent the fine, polished opinions of the BBC!
Shitification—yes, that’s a thing now! It’s the slow, painful, infuriating decline of services we once loved and depended on, now reduced to hollow shells of their former glory. Enter: Google.
First, they murdered Google Cache. Once upon a time, you could view a saved copy of a website that’s vanished from the web. And now? That magical button has gone to the place where hopes and dreams die. Why? Because Google said, “You don’t need it, darling!”
But wait, it gets worse. How many search results do you get now? 300. Yes, just 300 out of possibly millions! So unless you’re looking for Disney’s Mickey Mouse, your obscure little website is living in Google’s black hole, probably with those old Yahoo! email accounts we all forgot about.
And don’t even get us started on reverse image search. Once a glorious feature, now bastardized by Google Lens, which, let’s be honest, is just a glorified game of “What’s that blurry thing in your photo?” It’s useful for identifying dog breeds, but can be terrible for hardcore OSINT researchers like us who want to verify images based on sites and dates.
Google is a bit like a high-end coffee shop that suddenly decided to serve instant. You still go there, but you’re deeply disappointed. Shitification has arrived, folks, and Google’s the poster child. Paul and me discuss possible solutions. Enjoy!


Get full access to Digital Digging with Henk van Ess at www.digitaldigging.org/subscribe