yes, as in singular one. Back in April 2026 Anthropic caused a lot of media noise when they concluded that their new AI model Mythos is dangerously good at finding security flaws in source code. Apparently Mythos was so good at this that Anthropic would not release this model to the public yet but instead … Continue…
...or how to make Qwen learn query optimization via agentic reinforcement learning
Thirty years ago, Brewster Kahle founded the Internet Archive with an ambitious goal: Universal Access to All Knowledge. Today, that mission continues to grow with an exciting new chapter: the launch of the Internet Archive Switzerland, a non-profit foundation based in St. Gallen. The Internet Archive Switzerland…
We shared the next step in our journey to bring together the best of a search engine with the best of AI.
The surprisingly difficult ordeal of sending a decade-old MacBook to a friend in a Ugandan refugee camp.
Abusing an unauthenticated Bluetooth protocol to turn a PC speaker into a Rubber Ducky.
What if you could strap a full desktop GPU to your MacBook Air? Turns out, you can.
Watermarks that spy on users are no mere watermarks
I had a wierd experience which made me realize that the internet is probably dead.
Every so often the Earth produces a signal that is impossible to ignore. This graph is one of them. It shows sea-surface temperatures in the Niño 3.4 region of the equatorial Pacific, one of the most important parts of the Earth's climate system. Each blue line represents a different year since 1982. The red line is…
GLM 5.2 is the first open weights model I'd call a genuine competitor to Opus and GPT for agentic work - at ~15-20% of the price. Part one of why AI inference margins are about to collapse.
Bug Bounty Hunter & Researcher
I’m part of the desktop software generation. From Windows 95 to Windows 7, I grew using mostly-offline software on computers operated via mouse and keyboard, well before tablets and smartphones. Recently, I’ve been missing one particular part of that era: its consistency in design. I want to tell you about
California Attorney General Rob Bonta alleged Amazon led a price-fixing conspiracy of online retailers to inflate prices across the entire economy. And he's demanding an immediate halt.
xAI is renting huge amounts of GPU capacity to Anthropic and Google. Financial engineering ahead of the SpaceX IPO, a real compute shortage, or a genuine datacentre advantage? Probably all three.
We have the whole Oxide team coming to Emeryville this week for our annual OxCon meetup. For a remote company, an in-person meeting is uniquely energizing, and in preparation for that, we made some t-shirts featuring Oxide logos that are an homage to past computer companies.
Everyone is talking about Jev - here it is in 25 lines of Python.
For some reason or other, people have been posting a lot of excerpts from old emails on Twitter over the last few days. The most vital question everybody’s asking themselves is: What’s …
Who were the most popular personal bloggers of 2025, and what made them successful on Hacker News?
Oh, you use Firefox? Name every extension.
What would you do if your organization had used a domain name for 27 years, and the registrar holding the domain seized it without any advance warning? All
This one is about my friend, Aaron. I met him at school, and in our first year when we were eleven years old I told him I wanted to be his best friend but he…
Apple should end their prohibition on shapes in MacOS app icons
Agents made implementation cheap, so I started doing far more of it in parallel. My GitHub history since 2019 shows what that did: as VP of Technology I am back at my full-stack engineer pace, with bursts, burnout and a low season that is no longer low.
VoidZero, the team behind Vite, Vitest, Rolldown, Oxc, and Vite+, is joining Cloudflare. Vite stays open source, vendor-agnostic, and built for everyone.
How an expired nameserver let me take over e164.arpa zones for multiple territories, and why I probably should have checked my logs sooner.
Pi now supports MCP. Why we changed our minds, what changed in MCP and in Pi, and what Codemode is.
An overview of how Multi-Token Prediction (MTP) drafters are making Gemma 4 models up to 3x faster at inference.
When AI handles 95% of your incident response, do you get worse at handling the 5% that actually matters?
