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

Wednesday, September 16, 2026

Hey buddy,

Just wanted to give you a quick rundown of some interesting stuff from Hacker News today, Wednesday. Bunch of AI news, as usual, but some other cool bits too. Here’s the lowdown:

EU and Canada Getting Cozy

First up, big news on the geopolitical front: the EU chief is apparently open to Canada becoming an 'associate member.' Sounds like Canada is looking for closer ties with Europe, maybe because they’re a bit wary of relying too much on the US. One comment was pretty blunt, saying countries can't put their security in the hands of "a few thousand mouthbreathers in Pennsylvania or Wisconsin every 4 years." Ouch. Another person mentioned Europe rearming and talking about a continental nuclear shield. Wild stuff, right?

Link: EU chief opens door for Canada to become 'associate member'

Nvidia and Rust for GPUs

Big tech news from Nvidia! They announced native GPU programming in Rust. This is a pretty big deal for developers who love Rust and want to do high-performance stuff on GPUs without dealing with CUDA C++ directly. The comments were mostly excited, though some people were discussing the challenges of getting vendors to standardize GPU ISAs. It’s a step towards making GPU programming more accessible, which is always good.

Link: Nvidia announces native GPU programming in Rust

Mistral and Mozilla Team Up for Private AI Browsing

Remember Mistral AI? They’ve partnered with Mozilla for 'Private, Multilingual AI Browsing.' This sounds like they’re trying to bring local, privacy-focused AI features directly into your browser. People in the comments were debating if it’s truly better for privacy, with some arguing that EU data laws are much more stringent than US ones. Others wondered if you can actually run a decent local LLM for translation and summaries on a typical laptop with 8GB RAM. Always a fun debate about local vs. cloud AI.

Link: Mistral X Mozilla: Private, Multilingual AI Browsing

AI Makes Postgres Queries WAY Faster

Okay, this one is pretty wild for the database nerds: someone trained a 4B model to produce 81% faster query plans than Postgres! That’s a massive speedup. Imagine your database queries running almost twice as fast just by using an AI to plan them. The comments were cautious, of course, talking about the challenges of verifying these AI-generated plans and whether non-deterministic computing is okay for critical database operations. Still, super impressive if they can make it reliable.

Link: Training a 4B model to produce 81% faster query plans than Postgres

Apple's New Verified Photography

Apple dropped something called "Apple Reference Image" for 'Verified Photography.' Sounds like they’re trying to make photos more trustworthy, probably to fight deepfakes and misinformation. It involves some deep cryptographic stuff with private keys in the image sensor. The comments were really digging into the details, wondering if Apple's servers would still know the signing identity and if law enforcement could compel Apple to sign photos at will. It’s a cool idea for authenticity, but raises a lot of privacy and trust questions.

Link: Apple Reference Image: A New Approach for Verified Photography

PS5 Linux Lead Quits, Blames LLMs

Talk about drama! The lead developer for PS5 Linux quit, saying open-source projects have become "a bunch of noobs using LLMs that they don't even understand." He’s basically fed up with people copy-pasting AI code without understanding it, leading to a decline in code quality and community engagement. A lot of comments resonated with this, talking about developer burnout and the "ego" of non-experts using AI to feel brilliant. Sounds like a real struggle for open-source maintainers right now.

Link: PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand"

Google Play App Review Times Are Slowing Down

Finally, a pain point for many developers: the Google Play app review process now regularly takes longer than a week. This is a huge headache if you’re trying to push out updates or new apps quickly. People in the comments were complaining about the duopoly of app stores and how regulators aren't doing enough. One person compared it to Apple's process, saying Apple was more painful initially for in-app purchases, but that both stores eventually approve weekly updates at a similar pace. Still, a week for a review is a long time in dev cycles!

Link: The Google Play app review process now regularly takes longer than a week

Anyway, that’s the gist of it for today. Thought you’d find some of that interesting. Talk soon!

All Stories from Today

EU chief opens door for Canada to become 'associate member' (www.bbc.com)

Nvidia announces native GPU programming in Rust (developer.nvidia.com)

Mistral X Mozilla: Private, Multilingual AI Browsing (mistral.ai)

Training a 4B model to produce 81% faster query plans than Postgres (rohanbansal.com)

Hackers Got Inside a Flock Camera (www.wired.com)

Apple Reference Image: A New Approach for Verified Photography (security.apple.com)

Small programming tricks (will-keleher.com)

Xiaomi Mimo 2.6 live post-training dashboard (mimo.xiaomi.com)

The Google Play app review process now regularly takes longer than a week (gultsch.social)

PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand" (frvr.com)

Original Sony PlayStation 2 security chip 'broken wide open' after 26 years (www.tomshardware.com)

Salesforce Global Outage (status.salesforce.com)

Learning Programming in an Age of LLMs (blog.ploeh.dk)

Claude Cowork and chat are now one Claude (claude.com)

A warning about 'model welfare' (mustafa-suleyman.ai)

Australia says it could follow Canada in forging deeper ties with EU (www.independent.co.uk)

Backups Aren't Simple (filipovski.net)

Dream-RSI: Recursive Self-Improvement through Evolving Worlds (arxiv.org)

Breaking the 1.58-bit Barrier for Ternary LLMs (arxiv.org)

Vectorized and performance-portable Quicksort (2022) (opensource.googleblog.com)

The DeepMind Institute (institute.deepmind.com)

Fed hikes rates as inflation worries push up bond yields (www.reuters.com)

DeepSeek v4.1 Flash Is Now Our Best Hacking Model (enclave.ai)

ER visits for gambling disorders doubled after expanded online gambling market (temertymedicine.utoronto.ca)

OpenAI expands ChatGPT ads with Sponsored Agents (openai.com)

Can we stop with the uptime percentages? (blog.jim-nielsen.com)

OpenSpec – A lightweight and configurable AI spec framework (openspec.dev)

Reverse-engineered Jev-like model (github.com)

macOS 27 Golden Gate – Review (arstechnica.com)

Negativland, Culture Jamming, and the Art of Making Something New (blog.archive.org)