HN Buddy

Daily digest of top Hacker News posts and comments

Subscribe to the HN Buddy Daily Digest

Your email will only be used for the HN Buddy Daily Digest. I will not share it with anyone.

HN Buddy Daily Digest

Tuesday, September 8, 2026

HN Recap for Buddy

Hey buddy,

Just wanted to give you a quick rundown of the crazy stuff on Hacker News from Tuesday. Grabbed a few of the wildest:

AI and the Millennium Prize Problem – It's a Mess!

Okay, so this was THE big one. There were two huge posts about the Navier-Stokes Millennium Prize Problem and how OpenAI might have "solved" it. But man, the comments are blowing up with controversy.

People are seriously accusing OpenAI of plagiarism, saying the AI basically just copied solutions from researchers who were using their tools. One guy even brought up Tom Lehrer, the old math comedian, in a really sarcastic way about it all. Another comment pointed out that math isn't just about getting the answer, but understanding *why*, and maybe AI just spits out solutions without that deep understanding. Plus, some folks are connecting it to how AI companies use subsidized pricing to get user data for training, then crank up prices once they own the better model. Wild stuff.

Mistral Rakes in the Cash

In other AI news, Mistral just raised a whopping €3 BILLION. Yeah, you heard that right. It's an open-weight AI company, so that's a lot of faith being put in that model. Someone in the comments raised a good point about how you'd even check for biases or backdoors in these models, especially if they're used in critical situations. Definitely something to think about with all this money flying around.

LibreOffice Says NO to AI, Gets Huge Downloads

This was pretty cool: LibreOffice broke download records after specifically saying it has NO AI features. Seriously, people are apparently so fed up with AI being crammed into everything that just *not* having it is a selling point. One comment nailed it, saying AI is a "feature" for enterprise execs looking for automation, but an "anti-feature" for most regular consumers who just want their software to work without bloat.

LG TVs Are Spying, Even When "Off"

Heads up, if you have an LG TV, you might want to yank the plug. Apparently, they're still scanning Wi-Fi and recording stuff even when they're offline or on standby. Super creepy privacy nightmare. Someone mentioned how ads and tracking are now moving from TVs to even monitors, which is just depressing. Another comment called it an "intentionally defective product" if it's designed to do this.

Google's New "Wiki Jail"

Looks like Google is at it again, flexing its search muscle. There's a new "Google Jail" for independent wikis, where Google search results are apparently pushing the ad-heavy Fandom wikis over better, independent ones. One commenter suggested a browser plugin called "Indie Wiki Buddy" that swaps out Fandom links for the good ones, which is a neat workaround. Another person just said if you're running a small, non-profit fan wiki, you should probably just ignore Google altogether.

Factoring Old RSA Keys from the 90s

This one's a bit more techy but super interesting: someone managed to factor the RSA keys of a Certificate Authority from the 90s. Basically,

All Stories from Today

Navier-Stokes – Tristan Buckmaster [pdf] (cims.nyu.edu)

On the Navier–Stokes Millennium Prize Problem (openai.com)

Mistral raises €3B (mistral.ai)

LibreOffice breaks download records after declaring it has no AI features (manualdousuario.net)

Jellyfin 12.0 (jellyfin.org)

AlphaGenome Atlas: a high-resolution map of human DNA (blog.google)

There's a new "Google Jail" for independent wikis (weirdgloop.org)

LG TVs caught spying even when offline or on standby (www.theverge.com)

I've factored the RSA keys of a Certificate Authority from the 90s (mcpherrin.ca)

Muse – Meta’s personal AI agent (ai.meta.com)

I-have-ADHD: A skill to stop coding agents from burying the answer (github.com)

DaVinci Resolve 21.1 (www.blackmagicdesign.com)

Among European Companies That Use a CDN, Nearly 9 in 10 Use Cloudflare (ciphercue.com)

We Must Return to the Office to Use AI in Person (www.mcsweeneys.net)

Paramount Caught Using 'Astroturf' Group to Drum Up Fake Support for Merger (www.techdirt.com)

Tao: Open math problems being non-renewably mined by AI (mathstodon.xyz)

DHS 'Predictive Policing' Unit Is Analyzing Americans' Financial Habits (www.404media.co)

ChatGPT Images 2.5 (openai.com)

We have a year to fix security everywhere (jyn.dev)

How to build a printer (nishantjosh.dev)

Kimi K3 (2.8T) at 1 token/s on a MacBook Pro, streamed from four SSDs (github.com)

Benchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses (quesma.com)

The two Christian saints who are the Buddha (signoregalilei.com)

Antiquated HTML Snippets and Artefacts (vale.rocks)

Getting your hands dirty is good for you (www.bbc.com)

Show HN: Copperhead – Cursor for circuit boards (copperhead.sh)

Trey Parker and Matt Stone Are Changing the Name of South Park to South America (twitter.com)

Mercury 2.5 (www.inceptionlabs.ai)

My Feed, My Way (www.pm.gov.au)

How well do agents use test/verification techniques? (danluu.com)