#28 Summer kicks off with its scorching blaze
who will be reduced to ashes first?
As of this Tuesday, it's officially summer. It was already horribly hot, but now we can complain with full rights. And as if the high temperatures weren't enough, we're witnessing nations playing tug-of-war with nuclear devices. We're literally one slip away from someone mistaking the coffee machine button for something else (cue the recommended scene from Monsters vs Aliens).
Today's fiercest headlines and the tech world seem to rhyme more than ever. We're no longer asking ourselves if AI will claim victims in some economic sector. No — now we're wondering which profession will be the first to go extinct, like those elevator buttons you see in certain films. If it's a Coen brothers movie, the bellhop is always named “Chet.”
There’s been talk about radiologists, psychologists, teachers, lawyers… We'd like to throw two ideas into the debate:
1. At least in Spain, and defying all logic, nobody has managed to take down the court representatives.
2. Meta has taken aim at advertising agencies: starting in 2027, it neither needs nor wants intermediaries.
You’re reading ZERO PARTY DATA. The newsletter about current events, technopolies, and law by Jorge García Herrero and Darío López Rincón.
In the spare time this newsletter leaves us, we like to tackle tricky problems in personal data protection. If you’ve got one of those, give us a little wave. Or contact us by email at jgh(at)jorgegarciaherrero.com
Thank you for reading Zero Party Data! Sign up!
🗞️News from the DataWorld 🌍
.– If this next piece of news hadn’t made it into the DataWorld section until now — hell, not just made it in: capitalized it — it’s only because, naïve me, I simply hadn’t believed it: Microsoft cut off email access to Karim Khan, Chief Prosecutor at the International Criminal Court in The Hague, following one of Trump’s daily tantrums.
Well, despite what your supposedly rational brain might tell you, this wasn’t fake news. We read it on Politico. Twice.
Our take is dark, realistic, and threatening. And the title of the latest column on the matter says it all, loud and clear: Trump can pull the plug on the internet, and Europe can't do anything about it.
.– USA is taking social media screening a step further upon entry: if you’re applying for a student visa, you’ll soon be required to make your social media profiles “public” and hand over usernames and other details from the past 5 years. A full-on background check of your online life — just because Dorito. Via Jorge Morell. Dark times that still somehow keep getting darker.
.– Does the phrase “stochastic parrots” ring a bell when it comes to LLMs? It was coined by Emily Bender, and this venerable lady has a book coming out and recently gave an entertaining interview to the Financial Times. Well, I’m here to tell you that one of her quotes, this one…
… ties right into the bottom line of last week’s hilarious episode of John Oliver’s show, about AI slop — which I can’t recommend enough.
.– EDRi has told the European Commission that it’s time to reevaluate Israel’s GDPR adequacy decision. What’s happening now makes the issue even more obvious and pressing, but there were already serious doubts before about whether Israel really met the requirements for such a decision. What we know about PEGASUS certainly doesn’t help argue that there’s no undue access or carte blanche for its authorities.
Let’s not forget the European Commission has a habit of approving adequacy decisions based on outdated assumptions — often still referencing Directive 95/46/EC and national laws from third countries that no longer even exist. Just kicking the can down the road, as usual.
.– This site compiles a series of irregularities at OpenAI that will surely make for a compelling book someday. The conflict-of-interest section alone made the former corporate lawyer in me want to tear out what little hair I have left.
📄High density docs for data junkies☕️
.– A new UK regulation is coming that could put the adequacy decision at risk. As we mentioned here before, an extension was granted in anticipation that time would be needed to assess the potentially dangerous “Data (Use and Access) Act 2025 (DUAA).” The ICO has released a repository of documents on the topic: summaries of what it means for businesses, law enforcement, how it will affect citizens, and “a detailed summary of the changes for data protection experts.”
.– How do you audit an algorithm? One way is by thoroughly analyzing its outputs. That’s exactly what the authors of Not Even Nice Work If You Can Get It: A Longitudinal Study of Uber's Algorithmic Pay and Pricing (Reuben Binns, Jake Stein, Siddhartha Datta, Max Van Kleek, Nigel Shadbolt) have done. To no one’s surprise: riders pay more and more, drivers earn less and less, and Uber walks away with the lion’s share.
💀Death by Meme🤣
🤖NoRobots.txt o the AI stuff
.– The document was published last week, but it’s a good reminder that AI is also being positioned for a new Cold War among military powers, for all their espionage, counterespionage, and whoever gets caught in the middle. The Chinese aren’t lagging behind — they’re already putting their homegrown DeepSeek to work for such purposes. This report on how China is sticking its toe into developing Generative AI (GAI) for military use (the usual “national security” justification) is quite interesting. Naturally, they’re training the model with that classic capitalist-greed bias versus revolutionary-comrade framing.
New concept unlocked: “cognitive warfare.” Apparently, “disinformation” wasn’t dramatic enough anymore.
.– Academic research is now beginning to seriously ask whether excessive AI use is dumbing us down. There’s a big difference between using AI as a tool under your control and just letting it replace you — like that politician in Panama who read aloud everything Chat had written for him, including the part where it asked if he wanted more info on a specific topic.
.– Anthropic gave Claude 4 access to emails that directly revealed (i) that there were plans to shut it down and replace it with another model, and (ii) that the engineer involved was having an extramarital affair. Claude wasted no time — it immediately blackmailed the allegedly hot-blooded user.
Naturally, this sparked a lot of attention. Anthropic’s comment: “It wasn’t just Claude — all frontier models did the same.” Not exactly reassuring. At least not to me.
Via Simon Willison. Nothing more to add — game on.
📃Paper ot the week
.– This is the one: Sharing Trustworthy AI Models with Privacy-Enhancing Technologies, by Christian Reimsbach Kounatze — which we found via Katharina Koerner.
🧷Useful tools 🔧
.– This German website processes a couple of files from your Twitter portability file and can tell you exactly which personalized ads were shoved your way.
🙄 Da-Ta-Dum Bass!
.- We never disappoint as a species. Here’s a sample of the wonderfully secure passwords most used this year. The rest of these brilliant, ironclad choices can be found in the post by the folks who carried out this thrilling study.
If you think someone might like—or even find this newsletter useful—feel free to forward it.
If you miss any document, comment, or bit of nonsense that clearly should have been included in this week’s Zero Party Data, write to us or leave a comment and we’ll consider it for the next edition.