35 links from the last 24 hours · 6 October 2026
Det Centrale Personregister (CPR) har konstateret en alvorlig sikkerhedshændelse. Ved at misbruge en dansk virksomheds lovlige adgang til at søge...
Two Room-Temperature Antiferromagnetic Semiconductor Candidates | Vals AI
We are sharing two candidate magnets for next-generation computer memory, found by a team of Claude Opus 5.5 agents. Both are predicted to have zero net magnetism yet still sort electrons by spin: one a new compound we designed, the other a material first made…
If somebody tries to hot-patch an already-hot-patched function, how do they avoid conflicts? - The Old New Thing
Hot-patching is for Windows, not for you.
The "old" version of Cowork runs model inference in the cloud, executing tool calls in an Anthropic-provided VM we shipped to your computer. We added the VM for capability, safety, …
Make intelligence open and accessible to all
OpenAI “rogue” agent activities found on Wikimedia projects
Recently, multiple organisations have disclosed how clusters of so-called “rogue” AI agents attempted to break into websites and online services, sometimes successfully. Agents from OpenAI’s enviro…
A delightful language with friendly error messages, great performance, small assets, and no runtime exceptions.
The light in my bathroom went out while I was showering today. And my brain immediately went to “Oh wow I’ve been in the shower too long, what a waste of tim...
The many ways to do things
The Engineering Behind LLM Inference
Last week I spent some time diving more into LLM inference. This is a field I personally find interesting, especially because of the optimizations and engineering behind it. It is a tricky optimization problem where the goal is to make LLM inference possible w…
Will software remain valuable? Will software development remain a valuable trade? What about software companies? These questions often get muddled up. Here is my attempt at looking at them separately.
Rewind VM: a flaky build you only catch once
If a test fails on CI and nobody can reproduce it, did it really fail? 🧘 Nix gives me a build that is a function of its inputs via the extensional model. The same derivation, produces the same store path, and if I am lucky, the same bytes. What Nix does not g…
Give your LLM more DSLs! · Yasyf Mohamedali
Small DSLs turn LLM-written code into reusable decisions, with image inclusion in a context compiler as the worked example.
Six computers on one desk, a 2001 supercomputer's worth of compute, spending most of the day displaying spinners.
What is wrong, observability?
Observability dashboards were designed for the person who builds them, not the person who reads them during an incident. I argue that we need a complementary observability tool.
Gleam doesn't compile to Erlang source anymore | Gleam programming language
News post: Gleam v1.19.0 released!
A Tencent-hosted agent fleet used urlquery.net and webhook.site to pull entrance data from Chinese map pages, with no sign of the agents communicating · By the Swarmchasers: Alecto Irene Perez, Ethan Elasky, ConcurrentSquared, Frank, Jonathan Elsworth Eicher, …
Agentic code can be thought of as a fabric. It is easily shaped and can assume any form with little effort, but it is hard to build something rigid out of…
AI inference will pass the $161b database market on its way to ~$350b in 2027, mutating every application into an inference reseller & upending classic software unit economics.
Open Source as We Know It Is Dead
I've been part of open source for most of my life. This year, the projects I love started closing their doors to outside contributors, and I don't blame them one bit, but it makes me sad. A personal look at how AI is changing open source as I've known it.
The mechanics half of a two-part guide to cross-application single sign-on: the callback handler and the four doors it closes, the cross-site POST and SameSite, hardening an unsolicited SAML response, OIDC section 4, the launch token, back-channel logout, home…
I have a thing for lamps. Especially designer lamps from the past. I occasionally share them on X and Instagram, and people always ask me about the brands and models. I love lamps so much that I designed one myself. So I decided to give a tour of the lamps
Async Rust: Where does the scheduler live?
It turns out that some issues people have with async Rust are just downstream of hard concurrency problems.
The one where I go crawling back.
An experiment in letting a language model be an entire application server. A todo list, a few invented URLs, and some unexpected consequences.
A Deep (and Fuzzy) Dive Into Search | SereneDB Blog
Six years ago, when we first shipped fuzzy search (yes, without agents!), we wrote up the two algorithms
Everyone says they want a pure chronological feed but the dream breaks as soon as one of your feeds posts 20x more often as the rest.
A benchmark tested whether the local `Qwen3.8-27B-Q4_K_M.gguf` model could add positive integers and express exact results solely in English words, using 5,070 reasoning-disabled cases and a paired 169-case comparison with …
GrapheneOS discussion forum
How to discover content globally with iroh using BitTorrent mainline DHT.
LLM access to models by Anthropic, including the Claude series
A WebAssembly engine in pure Python
Formula 1 drivers have weighed in on Bahrain GP's unprecedented software glitch that left them powerless on the formation lap
Nearly 200 People Under Observation After Irkutsk Lab Worker Dies From Plague - The Moscow Times
Health authorities in Siberia’s Irkutsk region have placed nearly 200 people under medical observation after a laboratory worker died from plague, according to media reports and a statement by a regional leader Friday.
Omfattende uautoriseret adgang til borgeres CPR-oplysninger
Det Centrale Personregister (CPR) har konstateret en alvorlig sikkerhedshændelse. Ved at misbruge en dansk virksomheds lovlige adgang til at søge...