what i read this week - week 23 2026
I've read less this week than the prior one. I was catching up on a lot of recorded talks from re:publica 26, which I couldn't attend (those talks will get their own post!). I was also a tad busy with meeting people and doing the ring class, and also unfortunately struggling with depression. I didn't manage reading a lot of the stuff I meant to read. Oh well.
Here goes!
online articles
Character.AI enshittification - good example of what will arrive at other models and use cases soon, inevitably, especially since this stuff is so expensive.
EU's Digitalisation Push: Surveillance, control, exclusion - Made me learn about the concept of the increasingly "digital welfare state" and how governments worldwide are using tech to automate, predict, and "optimise the delivery of social protections". This means people most vulnerable, who need state safety nets the most, are also most at risk of having their privacy violated, their data collected and sold, and their behavior automatically analyzed. It's getting easier and easier for the EU to grab control over social security aspects instead of leaving it to the MS, due to classifying the digital exchange of data between governments, digitalization etc. as an EU issue. I have some issue with the presentation of the facts because they make it seem as if your data is only now just coming online, when in fact, it has been on government agencies' servers for over a decade already, and exchanged between them; it was just not accessible online to you. It was a hassle for you to get anything digitally in return. So yes, while the risks are there, and there are new projects and developments, a lot of it is just finally making sure you get an easier access to your own data as well, and hopefully have to submit less paperwork. I can say that as an employee in the public sector who was partially involved in our internal changes we made to comply with Germany's new Onlinezugangsgesetz (OZG; "online access act"). Our servers and databases stayed the same; now you just no longer need to send out a pigeon with a note and hope to hear something back within the next 5 years when you want something from [my employer]. They also underestimate how much need-to-know basis there is in granting access; no one will have a complete 360-view access to a digital life file. I don't like how it's just presented as inevitable that your doctor can see info about your housing situation, as if this would be in any way complying with any privacy regulation, or the reality of the software state of any doctor's office (it is bad). Aside from the digital aspect, the risk of the government just completely shutting off your lifeline has not increased, since they could already have all the info and decide to do so when your life was a folder in an office.
Linked in, and related to, the prior one: UN Digital Welfare Dystopia - from 2019, exploring how digital welfare is often used to cut welfare spending, set up intrusive government surveillance and generate profits for private companies as these supply the systems.
EU €200 Million Temu Fine - Temu breached the Digital Services Act as it failed to properly assess the systemic risks posed by illegal and unsafe products being sold in the EU.
The attack on competence - amazing read about how uses of LLMs get differently scrutinized at work, and how real competence (which is usually about constraints, structure, patience, testing, knowing your craft etc.) is outshined by this sort of fake competence companies value nowadays, in which the vibecoder who cares about none of these things is pushing out a lot of stuff and is seen as more impressive or working harder. They are protected and encouraged by the investor class (usually the management) to continue to use LLMs this sloppy way under the guise of "moving fast and breaking things", because they only care about the superficial numbers as it helps them build legitimacy by posting that stuff on LinkedIn or when they are speaking about technology at conferences and meetings. Management can begin to feel quite threatened by real competence, as the competent people are not yes-men like the vibecoders. AI promised this investor class to finally eliminate the need for competence and not have to consider or respect any feedback other than their own (which they generate with sycophantic AI). I'm... actually fighting a somewhat similar fight at work right now, so this hits home.
Debate over UK Migrant Age Assessment - AI wrongly assigns minors as adults, which poses dangers.
European operators get more satellite spectrum - European companies get prioritized in the satellite spectrum.
Orbans Payroll - The Lajos Batthyány Foundation and its Danube Institute spent a ton of money to buy influence on the Western Right, particularly spending it on public figures of the Western right-wing media and think-tanks.
Meta Silently Added Face-Recognition Code for Its Smart Glasses to Millions of Phones - (Archive) - an AI companion app by Meta that is needed to use key feature of the glasses has code for NameTag included now. NameTag will use biometric features of the people you're looking at to check if they are already registered on your phone, and for info/notifications regarding them. I assume this is for when you meet people you know and it shows you facts around them, a summary of your last exchange, etc.? I understand why people might think this is useful in a vacuum/ideal world, but damn, not like this, in this climate...
Why the AI Policy Debate Should Focus More on the Harness and Protocol Layers - loooong transcribed interview with the chief technology officer of Mozilla about AI, especially cybersecurity. I find his takes about the dual use nature a bit naive and evasive of the real issues, as well as how many issues AI causes for open source, but it was an interesting read. Shocking though that he spends $1,500 a month on his AI agent, and he can't even properly explain what he does with it, aside from... checking his calendar? But I am giving him grace in that sometimes you just aren't that great with explaining stuff spontaneously in a podcast. What I also don't like in general, in many formats and interviews like this one, how people can just say "We could build this very privacy-invasive tech in a way that is privacy-respecting, but we choose not to." which at first is right, depending on the tech, but then it's about Ring cameras, car cameras and other surveillance in public, and it's like - how? Either you get recorded and it alters your behavior, or you aren't. There is no middle way for some of this, but they always get away with saying that without offering specifics.
FAQ zur Bezahlkarte - German FAQ about the payment cards for refugees. It's a problem that every area handles them differently, support is patchy, they can not be used to pay online, teens cannot get one, and the cards allow a very low amount to be withdrawn and spent.
Gewerkschaftsgründung bei Wikimedia - German article about how Wikimedia (mainly their US branch) is attempting to unionize! <3 Please always unionize! Join your unions! I am in one, too, and I feel comfortable sharing that.
Gematik plant mehr Zentralisierung und Macht für sich - Organization responsible for our digital health stuff.
NRW will Haftung der Online-Plattformen erweitern - calls for platform owners to be held liable for deepfakes etc. under an amendment of the DSA. I think that makes sense, considering the decision impact of the Russmedia case.
press releases, policy briefs, drafts and activity reports
Recovery and Resilience Facility - official explanation of what the EU's Recovery and Resilience Facility (RRF) is. Member states are getting money for hitting specific goals in their plan, comprised of at least 37% going to green measures and 20% to digital measures. They must meet all milestones and targets by 31 August 2026, and the Commission must make the final payments by 31 December 2026.
Once Only - official explanation of the Once Only Principle -> federated IT architecture in Cross-border e-Services for Business Mobility, Updating Connected Company Data and Online Ship and Crew Certificates.
Single Digital Gateway - official explanation for the Single Digital Gateway -> facilitates online access to information, administrative procedures, and assistance services that EU citizens and businesses may need in another EU country.
Improving AI Labels - YouTube's new AI labels
Positionspapier Verbraucherzentrale zur Datenschutzaufsichtsreform - opinion of the German consumer rights organization on the coming changes to our data protection authorities. They make clear they want unambiguous responsibilities, same decision-making processes and results across the board (right now, each DPA is kinda giving their own, sometimes wildly varying positions), fast help and decisions for citizens, no centralization in the non-public sphere, a possible One-Stop-Shop on the national level, and more resources for our Datenschutzkonferenz (DSK).
Gesetzesentwurf zur Durchführung der Verordnung (EU) 2024/2847 (Cyberresilienz-Verordnung)
papers, studies and similar publications
None this time, couldn't make it. I have several open tabs of some though, especially around the new tech sovereignty package...
cases
- This one about how a parent using retail video surveillance footage in child custody proceedings is a legitimate interest under Article 6(1)(f) GDPR. I summarized it for GDPRhub.
books
I am also currently reading Arne Semsrott's new book; I am about 40 pages in (from a small ~180).
In total, that is roughly ~ 125 pages, if we count an article as two pages on average.
Reply via email
Published