31
Trapping misbehaving bots in an AI Labyrinth
How Cloudflare uses generative AI to slow down, confuse, and waste the resources of AI Crawlers and other bots that don’t respect “no crawl” directives.
Most cited
Posts that commenters keep linking to, ranked by how many different people have cited them in Hacker News comments. 296 posts.
32
Agentic Misalignment: How LLMs could be insider threats
New research on simulated blackmail, industrial espionage, and other misaligned behaviors in LLMs.
33
A 10x Faster TypeScript
Embarking on a native port of the existing TypeScript compiler and toolset to achieve a 10x performance speed-up.
34
Privacy Pass Authentication for Kagi Search
Today we are announcing a new privacy feature coming to Kagi Search. Privacy Pass is an authentication protocol first introduced by Davidson et al., in [1], and recently standardized by the IETF as RFCs [2—4]. Our...
35
Subliminal learning: Models transmit behaviors via hidden signals in data
We study subliminal learning , a surprising phenomenon where language models learn traits from model-generated data that is semantically unrelated to those traits. For example, a "student" model learns to prefer owls when trained on sequences of numbers generated by a "teacher" model that prefers owls. This same…
36
The Illusion of Thinking: Strengths and Limitations of Reasoning Models
Recent generations of frontier language models have introduced Large Reasoning Models (LRMs) that generate detailed thinking processes…
37
Data Centers and Water Consumption | Article
See all of our articles on the impacts of data centers , including their energy needs, water usage, noise pollution, and contribution to higher energy bills: The Environmental Impacts of Data Centers (Article Series)
38
Code execution with MCP for efficient agents
Learn how code execution with the Model Context Protocol enables agents to handle more tools while using fewer tokens, reducing context overhead by up to 98.7%.
39
My AI skeptic friends are all nuts
My smartest friends have bananas arguments about LLM coding.
40
Cloudflare outage on November 18, 2025 post mortem
Cloudflare suffered a service outage on November 18, 2025. The outage was triggered by a bug in generation logic for a Bot Management feature file causing many Cloudflare services to be affected.
41
Local-first software (2019)
A new generation of collaborative software that allows users to retain ownership of their data.
42
Sycophancy in GPT-4o
We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic.
43
Run a C# file directly using dotnet run app.cs
Run C# files instantly with dotnet run app.cs, no project file needed! Coming to .NET 10, try it out today in Preview 4.
44
Context Rot: How increasing input tokens impacts LLM performance
Large Language Models (LLMs) are typically presumed to process context uniformly—that is, the model should handle the 10,000th token just as reliably as the 100th. However, in practice, this assumption does not hold. We observe that model performance varies significantly as input length changes, even on simple tasks…
45
Effective learning: Rules of formulating knowledge (1999)
This article will help you overcome one of the greatest difficulties you will face when trying to accelerate learning: formulating knowledge.
46
Oklch in CSS: why we moved from RGB and HSL (2024)
CSS Color Module 4 adds oklch(), and we gain P3 wide-gamut support, boost code readability, and improve developer-designer communication.
47
NVIDIA Puts Grace Blackwell on Every Desk and at Every AI Developer’s Fingertips
CES—NVIDIA today unveiled NVIDIA® Project DIGITS, a personal AI supercomputer that provides AI researchers, data scientists and students worldwide with access to the power of the NVIDIA Grace ...
48
Android developer verification: Early access starts
News and insights on the Android platform, developer tools, and events.
49
Run DeepSeek R1 Dynamic 1.58-bit
DeepSeek R-1 is the most powerful open-source reasoning model that performs on par with OpenAI's o1 model. Run the 1.58-bit Dynamic GGUF version by Unsloth.
50
Using LLMs at Oxide
Large language models (LLMs) are an indisputable breakthrough of the last five years, potentially profoundly changing the way that we work. As with any extraordinarily powerful tool, LLM use has both promise and peril — and that they are so general-purpose leaves real questions about how and when they should be used…
51
Reasoning models don't always say what they think
Research from Anthropic on the faithfulness of AI models' chain-of-thought reasoning.
52
Circuit Tracing: Revealing Computational Graphs in Language Models (Anthropic)
We describe an approach to tracing the “step-by-step” computation involved when a model responds to a single prompt.
53
A letter from Sam & Jony
Building a family of AI products for everyone.
54
How to Build an Agent
Building a fully functional, code-editing agent in less than 400 lines.
55
Introducing GPT-5
We are introducing GPT‑5, our best AI system yet. GPT‑5 is a significant leap in intelligence over all our previous models, featuring state-of-the-art performance across coding, math, writing, health, visual perception, and more.
56
Perplexity is using stealth, undeclared crawlers to evade no-crawl directives
Perplexity is repeatedly modifying their user agent and changing IPs and ASNs to hide their crawling activity, in direct conflict with explicit no-crawl preferences expressed by websites.
57
The Llama 4 herd
We’re introducing Llama 4 Scout and Llama 4 Maverick, the first open-weight natively multimodal models with unprecedented context support and our first built using a mixture-of-experts (MoE) architecture.
58
Steve wants us to make the Macintosh boot faster
We always thought of the Macintosh as a fast computer, since its 68000 microprocessor was effectively 10 times faster than an Apple II, but our Achilles heel was the floppy disk. We had limited RAM, so it was often necessary to load data from the floppy, but there we were no faster than an Apple II. Once we had real…
59
A new layer of security for certified Android devices
Starting in 2026 and in select countries first, Android apps must be registered to a verified developer in order to be installed.
60
Octoverse: A new developer joins GitHub every second as AI leads TypeScript to #1
In this year’s Octoverse, we uncover how AI, agents, and typed languages are driving the biggest shifts in software development in more than a decade.