What I found interesting this time

“I Trained My Own AI… It Beat ChatGPT,” n.d. https://m.youtube.com/watch?v=aV4j5pXLP-I&pp=ygUdSSB0cmFpbmVkIG15IG9lbiBhaSBwZXdkaWVwaWXSBwkJvgoBhyohjO8%3D

PewDiePie just shown that AI benchmarks are trash XD
So much money was pumped into it. LOL.

“Open Source Devs Consider Making Hogs Pay for Every Download,” n.d. https://www.theregister.com/2026/02/28/open_source_opinion/.

“AI-Generated Art Can’t Be Copyrighted after Supreme Court Declines to Review the Rule,” n.d. https://www.theverge.com/policy/887678/supreme-court-ai-art-copyright

Supreme court from USA state that AI generated code can’t be copyrighted. Does that mean that when we hack companies with AI generated code it’s Free as a beer? :D

“You’Re Paying AI Companies a Monthly Subscription Fee to Be Fingerprinted like a Parolee.,” n.d. https://infosec.exchange/@k3ym0/116161635202253362

Short analysis on how much tracking is done by ChatGPT, Gemini and Claude.
tldr; even uBlockOrigin has problems with blocking the amount of traking noise made by these tools.
It’s really scary to think that people put there their sex preferences, details about their kids, medical records, or places where they live.
I think this can really backfire. Especially in countries like USA which do not respect any privacy measures. Higher insurance price. Online profiling based on what you shared with LLM provider. Deportation by ICE. It all sounds hard to imagine but not impossible.

“BullshitBench: Models Answering Nonsense Questions,” n.d. https://petergpt.github.io/bullshit-benchmark/viewer/index.html.

“Libraries That Quietly Revolutionized C,” n.d. https://youtu.be/kS_GqDp6IT4

Quick explanation on how STB libraries work. I find it amazing that if you don’t split your code into multiple files you can still achieve really maintainable code.

“Everyone in AI Is Building the Wrong Thing for the Same Reason,” n.d. https://www.joanwestenberg.com/everyone-in-ai-is-building-the-wrong-thing-for-the-same-reason/

Author is discussing problems of modern AI race like going without clear direction and building for sake of competition, not to deliver meaningful value to users.I recently looked as some YouTube videos where people were vibe coding websites and I must say the quality of the LLMs output got way better than it was 3 years ago. Sadly I still have not found anything better than example with one single HTML file.Whenever some AI "specialist" shows how they integrate AI workflow into bigger system there is a lot of "vibe" in it, but I have never seen the actual end product.