1
AI Capabilities 🧠
Source: original article
Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights We find that frontier LLMs are surprisingly good at natively reading and writing neural network weights , well past what we’d expect from humans or conventional algorithms. This might be a big deal for both optimization and interpretability. Just by looking at the raw weights, GPT-6 Astra can tell what a network was trained to do .
Actionable Insight
Frontier LLMs, exemplified by GPT-6 Astra, demonstrate an unexpected ability to natively read and write neural network weights. This capability allows them to understand a network's function and directly edit its parameters to improve performance, such as enhancing a small Qwen model's math skills. This breakthrough could significantly impact AI optimization and interpretability, offering new avenues for model development and understanding.
rikiyat.github.io
·
2790 pts
·
853 comments
·
by dbreunig
2
AI & Content Curation 🤖
Source: original article
It was Friday night at 11pm, I was tired, I went to YouTube to watch something and realized that I was looking at an enshittified wall of slop. Insanely exaggerated thumbnails, dumb recommendations of the same tired topics, truly terrible idiocracy-grade content. I'm sure this algorithm works well for the median person on the internet, but it definitely does not work for me. So this morning (Saturday) I took a hold of my Opus 5.5 and gave it the task: Let's write our own YouTube clone based on whatever videos have been recommended on Hacker News. And Opus came back with: hackertube.org !
Actionable Insight
HackerTube exemplifies how AI tools can rapidly create alternative content platforms, offering a curated experience distinct from mainstream "enshittified" feeds. This project underscores a user-driven demand for disaggregated content recommendations, potentially empowering individuals to build personalized media ecosystems. It suggests a future where AI facilitates the creation of niche platforms that challenge established content monopolies.
karpathy.bearblog.dev
·
2841 pts
·
632 comments
·
by karpathy
3
AI & Books 🤖📚
Source: Hacker News post
Saw Karpathy's HackerTube and wanted the same thing for books. So I had Claude build one (Opus 5.5). It's ~16.5k books that at least two different HN commenters brought up, from 2007 till now. Each book links back to the comments that mentioned it. Data is downloadable from the about page. To be clear: Claude wrote almost all of it, including the data pipeline and the site. I mostly steered and made calls, bought the domain name, etc. It counts mentions, so it's not perfect (praising a book and trashing it both count). If your favourite book is missing or matched to the wrong thing, tell me.
Actionable Insight
HackerBooks is an AI-generated index of approximately 16,500 books mentioned in Hacker News comments since 2007, with each entry linking back to its original mentions. The creator highlights that Claude (Opus 5.5) developed nearly all of the project, including the data pipeline and website, showcasing advanced AI capabilities in end-to-end project development. This initiative demonstrates an innovative approach to curating community-driven data using large language models.
hackerbooks.org
·
1828 pts
·
303 comments
·
by roansh
4
Security 🔒
Source: original article
In case you missed it: CVE-2026-98432 is a critical (CVSS 9.8)
bug in the Linux kernel's TCP code. It's been there since 2016, so virtually every Linux server on the planet
has it, and it can be hit remotely. Anthropic says it built a working exploit that gets root on a server. Debian patched it on October 16 . Ubuntu still doesn't have a patch more than two weeks later.
Actionable Insight
A critical Linux kernel TCP vulnerability, present since 2016 and remotely exploitable, remains unpatched in Ubuntu's default repositories weeks after disclosure and Debian's fix. Despite a CVSS score of 9.8, Ubuntu initially prioritized it as 'High,' leading to community concern over the delay and lack of transparent communication. This situation highlights the tension between rapid patching of severe vulnerabilities and the stability testing required for widespread distribution updates.
chalmovsky.com
·
1552 pts
·
647 comments
·
by chalmovsky
5
Software Development 🐛
Source: original article
Robert Haas: Postgres 19 is late, and it's AI's fault Over the past decade, the timing of PostgreSQL major releases has become very predictable: no earlier than the middle of September, and no later than the middle of October. This year is going to be different. PostgreSQL 19 most likely won’t be released until November, and it’s not impossible that the project could decide to postpone the release further. Moreover, this year, several rather large features were reverted quite late in the beta period: UPDATE/DELETE FOR PORTION OF , SQL Property Graph Queries (SQL/PGQ), and ALTER TABLE … MERGE PARTITION/SPLIT PARTITION were all reverted between August 27 and September 15.
Actionable Insight
The PostgreSQL 19 release is significantly delayed, with several major features reverted late in the beta cycle. This postponement is attributed to a substantial increase in bug discovery, largely facilitated by AI-assisted tools. The project faces the challenge of addressing a higher volume of identified issues, impacting its predictable release schedule.
rhaas.blogspot.com
·
829 pts
·
498 comments
·
by soheilpro
6
Wearable Tech 👓
Source: Hacker News post
Hi HN, we're Aarush and Arnav from The Subvocal Company (YC F26) ( https://subvocal.company/ ). We’re building smart glasses that let you talk to your computer without talking out loud. You can mouth words silently and discreetly and the glasses decode them to text at the speed of speech. Here's a video of Dave from YC trying it out: https://www.youtube.com/watch?v=9QdX1eObVYI (Edit: you don't actually have to move your lips or even open your mouth for this to work! Since the RF can see inside your mouth the system can be fine tuned on "internal speech". This isn't the automatic tongue movemen
Actionable Insight
Subvocal has developed smart glasses that convert internal speech into text, allowing users to interact with computers silently and discreetly. This technology bypasses the need for audible vocalization or even visible lip movement, offering a novel solution for hands-free computing and accessibility in environments where traditional voice dictation is impractical. Its ability to decode 'internal speech' represents a significant advancement in human-computer interface.
news.ycombinator.com
·
733 pts
·
348 comments
·
by aarushyadav
7
Cybersecurity Guides 🔒
Source: original article
NIST Has a Cybersecurity Guide Just for You | NIST Skip to main content Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.
Actionable Insight
NIST has released a cybersecurity guide aimed at self-employed individuals and solopreneurs. However, community feedback suggests the guide may miss its mark due to inaccessible language and jargon. Critics argue it lacks practical, actionable advice for its non-technical target audience.
nist.gov
·
252 pts
·
74 comments
·
by tomrod
8
Linux Development 🐧
Source: Hacker News post
I first released those in back in 2020 and now I finally added more architectures, Kernel selector, syscall descriptions and a few other niceties. It should work a bit better on mobile devices too but I am no CSS expert. The API is at https://api.syscall.sh
Actionable Insight
The updated Linux System Call website and API now offers expanded support for various architectures and kernel versions, alongside detailed syscall descriptions. This enhancement significantly improves its utility for developers working with low-level Linux programming. The added features make it a more comprehensive and accessible resource for understanding system call intricacies.
syscall.sh
·
225 pts
·
52 comments
·
by guitmz
9
AI Testing 🧪
Source: original article
AI QA Engineer & AI Testing Platform | Momentic Mo is an AI QA engineer that tests your app end to end. Verify critical flows and find bugs from a prompt. Trusted in production by teams shipping every day. "Our team's experience with Momentic has been great.
Actionable Insight
Momentic introduces an AI QA engineer, Mo, designed for end-to-end application testing to identify bugs and verify critical user flows. This platform leverages AI agents to interact with applications, aiming for high recall in bug detection. It allows teams to find bugs from a simple prompt.
momentic.ai
·
147 pts
·
98 comments
·
by wuweiweiwu
10
UI/UX ✨
Source: Hacker News post
I thought it would be fun to browse HN as if it were a newspaper. The site updates once an hour. Enjoy!
Actionable Insight
Hacker Times reimagines the Hacker News interface with a newspaper-inspired design, updating hourly to offer a different browsing experience. This approach aims to encourage users to engage with content more deeply than the traditional high-density HN layout. It reflects a broader community interest in alternative UIs for content discovery.
hackertimes.ca
·
331 pts
·
104 comments
·
by polpenn