1831
A small number of samples can poison LLMs of any size
Anthropic research on data poisoning attacks shows a few hundred malicious documents can "backdoor" large language models of any size.
New
The latest blog posts to take off on Hacker News. 6,328 posts.
1832
Figure 03, our 3rd generation humanoid robot
Introducing Figure 03
1833
Two things LLM coding agents are still bad at
I’ve been trying to slowly ease into using LLMs for coding help again lately (after quitting ), but something always feels off -- like we’re not quite on the...
1834
I played 1k hands of online poker and built a web app with Cursor AI
In the last two weeks I spent over a dozen hours playing poker, primarily online at pokerstarsmi.com, and live at a local casino. You can view the last 1,000 hands I played here: https://poker.rchase.com I spent at least as much time reviewing my hands with
1835
Now open for building: Introducing Gemini CLI extensions
Make Gemini CLI uniquely yours by connecting it to your everyday workflows and the tools you use most.
1836
Hacking the Humane AI Pin
On February 18, 2025, Humane announced it would be shutting down all services for the always online Humane Ai Pin in ten days, following a sale to HP . I had been vaguely aware of the Ai Pin, definitely more so than the average tech person, as I have a close friend who was one of the earliest adopters and eventually…
1837
Gemini 2.5 Computer Use model
Today we are releasing the Gemini 2.5 Computer Use model via the API, which outperforms leading alternatives at browser and mobile tasks.
1838
Less is more: Recursive reasoning with tiny networks
|| Paper | Code ||
1839
Vibe engineering
I feel like vibe coding is pretty well established now as covering the fast, loose and irresponsible way of building software with AI—entirely prompt-driven, and with no attention paid to …
1840
CodeMender: an AI agent for code security
Using advanced AI to fix critical software vulnerabilities
1841
When ChatGPT turns informant
The largely overlooked privacy risks of using AI apps that not only remember your conversations, but are capable of using these to reveal your deepest secrets to others
1842
A beginner's guide to deploying LLMs with AMD on Windows using PyTorch
Get started running LLMs with PyTorch on Windows using AMD consumer graphics hardware.
1843
Embracing the parallel coding agent lifestyle
For a while now I’ve been hearing from engineers who run multiple coding agents at once—firing up several Claude Code or Codex CLI instances at the same time, sometimes in …
1844
A History of Large Language Models
Large language models (LLMs) still feel a bit like magic to me. Of course, I understand the general machinery enough to know that they aren’t, but the gap between my outdated knowledge of the field and the state-of-the-art feels especially large right now. Things are moving fast. So six months ago, I decided to close…
1845
Why do LLMs freak out over the seahorse emoji?
Investigating the seahorse emoji doom loop using logitlens.
1846
What GPT-OSS leaks about OpenAI's training data
OpenAI recently released their open-weights model. Here we'll discuss how that inevitably leaks some information about their model training stack, and, on the way, show that GPT-5 was trained on phrases from adult websites....
1847
NIST's DeepSeek "evaluation" is a hit piece
How NIST Turned Open Science into a Security Scare
1848
1849
Managing context on the Claude Developer Platform
New context editing and memory tools enable Claude agents to handle long-running tasks without hitting context limits.
1850
1851
New antibiotic targets IBD and AI predicted how it would work
[...]Read More...
1852
Sora Update #1
We have been learning quickly from how people are using Sora and taking feedback from users, rightsholders, and other interested groups. We of course spent a lot of time discussing this before...
1853
How does gradient descent work?
This is the companion website for the paper Understanding Optimization in Deep Learning with Central Flows , published at ICLR 2025.
1854
OpenAI Is Just Another Boring, Desperate AI Startup
What is OpenAI? I realize you might say "a foundation model lab" or "the company that runs ChatGPT," but that doesn't really give the full picture of everything it’s promised, or claimed, or leaked that it was or would be. No, really, if you believe its leaks to
1855
Who needs Git when you have 1M context windows?
Lately I’ve heard a lot of stories of AI accidentally deleting entire codebases or wiping production databases . But in my case it was the other way around. I removed some working code and the LLM helped me to recover it.
1856
Which table format do LLMs understand best?
We benchmarked 11 data formats, including markdown tables, JSON, CSV, YAML, XML, and more, to see which LLMs understand best. We found surprising accuracy and token cost trade-offs.
1857
What makes 5% of AI agents work in production?
Beyond the Prompt: Notes from the Context Frontier
1858
Email immutability matters more in a world with AI
Email immutability matters even more in a world of AI. Here is how we enable your use of your data, in line with our values.
1859
How the AI Bubble Will Pop
The artificial intelligence boom is the most important economic story in the world. But the numbers just don't add up.
1860
