I'm joining Berget AI, an AI platform based in Sweden that offers sovereign AI infrastructure for European organizations.
I'll be focusing on developer experience and helping teams build with AI without compromising their data.
Are you evaluating AI infrastructure in Europe? Let's connect! π
23.02.2026 14:14
π 2
π 0
π¬ 0
π 0
Grafana dashboard showing ad blocking metrics, such as total and blocked queries.
Got around to replacing NextDNS with Blocky for my home lab.
Considered options like Pi-hole and AdGuard Home, but I kept coming back to the simplicity of Blocky and being able to version-control the entire configuration.
21.12.2025 07:33
π 0
π 0
π¬ 0
π 0
I love being able to export my configuration as text and just ask Claude whenever I mucked something up.
07.11.2025 19:23
π 1
π 0
π¬ 0
π 0
Got my first #Mikrotik router this week and Iβve honestly had so much fun migrating my home network (no, seriously!)
Been wanting to segment my IoT devices for a while, and I even managed to replace Tailscale for all my devices.
07.11.2025 19:23
π 0
π 0
π¬ 1
π 0
Last chance to turn it off.
On Monday, November 3rd, Microsoft will start using your LinkedIn data for AI training. And remember, you're opted in by default.
To toggle it off π Account - Settings & Privacy > Data privacy > Data for Generative AI Improvement.
31.10.2025 13:37
π 3873
π 3590
π¬ 89
π 209
The downside is I'd need to create announcement posts retroactively if I wanted to support comments for older blog posts.
27.10.2025 21:03
π 0
π 0
π¬ 0
π 0
Iβve seen similar solutions that source comments from GitHub issues, but I love the idea of using open protocols for this.
Of course, rate limits apply, but I should be so lucky to actually hit them π€£
27.10.2025 20:44
π 1
π 0
π¬ 1
π 0
Finally got around to adding a comments section to my blog posts, powered by Bluesky and Mastodon.
For each blog post, I create announcement posts on each platform and add the links to the posts in the front matter.
The replies on each post are then fetched client-side and interleaved.
27.10.2025 20:44
π 4
π 0
π¬ 1
π 0
Love to see in-person privacy controls π€
Personally, I donβt mind having my photo published, but I also want to relax and enjoy the event without having to think about cameras.
27.10.2025 08:33
π 1
π 0
π¬ 0
π 0
A plate with a seitan burger with home made fries, and a tall glass of lemonade.
In Berlin for @writethedocs.bsky.social π©πͺ See you all tomorrow!
26.10.2025 14:38
π 2
π 0
π¬ 0
π 0
Ever notice how #privacy controls and other opt-out mechanisms seem to have a curious proclivity for errors?
19.10.2025 06:48
π 1
π 1
π¬ 0
π 0
There's a special place in hell for teams that change the title of their website when you move to another tab.
Suddenly I have 10 tabs, all with the title "Please come back! π₯² ".
Today I even saw one that BLINKED! π€―
I don't care how great your website is, I'm closing your tab.
14.10.2025 11:26
π 1
π 0
π¬ 0
π 0
Piece of advice: While phrases like βX just killed Yβ might generate impressions, theyβre more likely to harm your credibility.
08.10.2025 10:44
π 1
π 0
π¬ 0
π 0
Westin's four states of privacy - Marcus Olsson
Learn about the foundational ideas by Alan Westin that helped define the privacy field.
Published: Intro to Westin's four states of privacy
I'm currently studying for the CIPT certification, and struggled to find resources on Alan Westin's work on the four states and functions of privacy. So I decided to expand my study notes into a blog post π
marcusolsson.dev/four-states-...
06.10.2025 10:55
π 1
π 0
π¬ 0
π 0
Love the new modern look! Looking forward seeing the other apps getting the same treatment.
30.09.2025 15:37
π 1
π 0
π¬ 0
π 0
Lumo is the first AI chat that is private by default.
Your chat history is protected with zero-access encryption, and Lumo keeps no logs, or paw traces. πΎ
πRead Lumo's Security Model β proton.me/blog/lumo-se...
06.08.2025 11:59
π 105
π 22
π¬ 24
π 5
Here's how longer-lasting smartphones will soon become a reality
New EU rules mean longer-lasting smartphones and tablets will soon become a reality as well as being more easily repairable.
Longerβlasting, repairable smartphones are now a thing in the EU! Since 20 June 2025, EU rules require smartphones and tablets to meet strict ecodesign and repairability standards, with mandatory labels showing battery life, charging cycles, repair-score, durability, and water/dust protection.
11.08.2025 07:45
π 7
π 6
π¬ 1
π 1
Introducing Lumo, the AI where every conversation is confidential | Proton
Lumo gives you the power to solve problems big and small, while keeping your personal data confidential. Try it now.
Proton has introduced Lumo, a privacy-first AI running on European data centers. All your chat conversations are end-to-end encrypted and Proton promises to never share your data or to use your data to train AI.
proton.me/blog/lumo-ai
25.07.2025 09:59
π 1
π 0
π¬ 1
π 0
Power move @cloudflare.social π
01.07.2025 21:25
π 0
π 0
π¬ 0
π 0
Open AIβs valuation is at $400 billion. They used about 190,000 books to train their plagiarism bot. Randomly saying 25% of their value comes from that stolen dataset, I think they owe the authors about $526,000 per stolen book.
27.05.2025 10:28
π 2604
π 847
π¬ 57
π 136
So wild to see those dipshits try to bully Zelensky, whoβs shown a level of bravery that theyβd be incapable of.
28.02.2025 18:59
π 28
π 1
π¬ 2
π 0
Ukraine is Europe!
We stand by Ukraine.
We will step up our support to Ukraine so that they can continue to fight back the agressor.
Today, it became clear that the free world needs a new leader. Itβs up to us, Europeans, to take this challenge.
28.02.2025 20:22
π 19341
π 5606
π¬ 669
π 521
OpenAI Says DeepSeek Used Its Work Without Permission to Create an AI That's Stealing Its Job
After spending years indiscriminately ripping off other people's work, OpenAI is trying to pin blame on Chinese AI startup DeepSeek.
somehow this is not an Onion article: OpenAI, which built its AI by stealing everything on the web with no credit, says that DeepSeek stole that pilfered work to make its own AI that's now eating OpenAI's lunch
futurism.com/openai-deeps...
29.01.2025 15:42
π 180
π 49
π¬ 5
π 21
If you are using Signal or other messaging apps: a gag order by the US government was just rendered ineffective, and Google and Apple have both confirmed that governments have been using the unencrypted contents + metadata of push notifications to spy on us!
So if you use Signal, for example, be sure to change this settings (other apps may have similar). Setting for Notification Content set to "No Name or Content".
The gag order was lifted in late 2023 - so I'm coming at this late, but this is the first I'm hearing of it, so sharing in case others didn't know.
So if you use Signal, for example, change the setting for Notification Contents to "No Name or Content". π§΅ /1
26.01.2025 15:29
π 112
π 72
π¬ 4
π 5
a woman says we were doing so well in a bravo ad
ALT: a woman says we were doing so well in a bravo ad
In Austin for the first time, and while there are many things I enjoy so far, Iβm struck by what I can only describe as a vegetarian minefield. Innocent looking salads with bacon fat, beautiful smoked cauliflower coated in anchovy.
15.01.2025 13:34
π 0
π 0
π¬ 0
π 0
Deploy Fides to Kubernetes - Marcus Olsson
Learn how to deploy Fidesβan open-source privacy-as-code platformβto a Kubernetes cluster.
Last year, I published a tutorial on how to get started with #Fides, an open-source privacy engineering platform.
For those who want to take the next step, my latest post covers how to deploy a minimal Fides installation to a Kubernetes cluster.
#Privacy #PrivacySky
10.01.2025 16:55
π 0
π 0
π¬ 0
π 0