You busy? Just wanted to quickly tell you about some of the wild stuff that popped up on Hacker News today. It was a pretty interesting Saturday!

HackerTube: YouTube built from videos recommended by HN

First up, you know how we're always seeing cool videos linked on HN? Well, someone actually went and built a whole YouTube-like site, called HackerTube, just from those recommendations. How cool is that for cutting through all the usual YouTube noise? In the comments, it was interesting to see one of the HN mods actually step in to remind people about keeping discussions civil, which is always a good thing.

Serbia just elected the world's brainiest parliament

Okay, get this – Serbia apparently just elected the 'brainiest parliament' in the world. Like, full of super-smart academics and scholars. But the comments were pretty split on it. A lot of people were like, "Yeah, but smart doesn't always mean good at governing," and that academics can be really socially awkward or just argue too much. So, not everyone's convinced it's automatically a good thing for the country!

TCP Use-After-Free: CVE-2026-98432 / CVSS 9.8

Then, on the more serious tech side, there's a pretty nasty security flaw in Linux, a TCP "use-after-free" bug with a super high severity score, like 9.8 out of 10. So, a big deal for server security. One comment brought up how Rust-based tools seem to have more trivial bugs compared to the deep, nasty logic bugs you often find in older C code, which is an interesting thought about language safety.

An Italian farmer found some unknown seeds in his attic and planted them

Switching gears completely – remember those old stories about ancient seeds? An Italian farmer just found some unknown seeds in his attic, planted them, and apparently grew some biblical-era grain. Super cool! The story sparked a bit of a debate in the comments about GM foods and their nutritional value versus these ancient varieties, and how some of these old grains have unique properties.

Tell HN: Scribd Breached

Another security hit: Scribd got breached. You know, for documents and stuff. People were talking about how virtual cards could really help prevent some of the payment data issues in breaches like this. And get this, "dang," one of the main HN admins, even jumped in on a comment thread, saying he's been reluctant to change some things on HN for various reasons. Pretty rare sighting of the man himself!

Anthropic blocks users for anything touching training larger models

On the AI front, Anthropic, that company making Claude, is apparently blocking users if they try to use their models to train *other* large AI models. Sounds like they're really trying to protect their turf and prevent competitors from leveraging their tech. Some users are finding Anthropic's "guardrails" super restrictive compared to what you can do with OpenAI's stuff. It's the classic "you don't own the platform" lesson again, but with cutting-edge AI services.

Do the math: How we took over ChatGPT accounts

And speaking of AI security, someone found a way to take over ChatGPT accounts! Pretty big deal, right? But the comments were kinda shocked at how little OpenAI apparently pays for these kinds of bug bounties – like, only 5 grand for a major exploit. Also, one of the commenters mentioned a tool that even flagged the report about the exploit as 100% AI-written, which is kinda meta, right? An AI reporting on an AI hack!

Anyway, that's the gist of the interesting stuff from today. Catch you later!