Posts that commenters keep linking to, ranked by how many different people have cited them in Hacker News comments. 131 posts.
We conduct a randomized controlled trial to understand how early-2025 AI tools affect the productivity of experienced open-source developers working on their own repositories. Surprisingly, we find that when developers use AI tools, they take 19% longer than without—AI makes them slower.
Anthropic's latest interpretability research: a new microscope to understand Claude's internal mechanisms
Reproducibility is a bedrock of scientific progress. However, it’s remarkably difficult to get reproducible results out of large language models. For example, you might observe that asking ChatGPT the same question multiple times provides different results. This by itself is not surprising, since getting a result from…
We investigate the internal mechanisms used by Claude 3.5 Haiku — Anthropic's lightweight production model — in a variety of contexts, using our circuit tracing methodology.
A technical report on three bugs that intermittently degraded Claude's responses: what happened, why fixes took time, and what we're changing.
We propose measuring AI performance in terms of the length of tasks AI agents can complete. We show that this metric has been consistently exponentially increasing over the past 6 years, with a doubling time of around 7 months. Extrapolating this trend predicts that, in under a decade, we will see AI agents that can…
Get more out of Claude Code: tell Claude what it can't see, give it a check to run, and enforce the rules that can't have exceptions.
Compare AI model performance on AA-Omniscience v1.1: Knowledge and Hallucination Benchmark. A benchmark measuring factual recall and hallucination across various economically relevant domains.
Anthropic research on data poisoning attacks shows a few hundred malicious documents can "backdoor" large language models of any size.
It turns out we've all been using MCP wrong. Most agents today use MCP by exposing the "tools" directly to the LLM. We tried something different: Convert the MCP tools into a TypeScript API, and then ask an LLM to write code that calls that API. The results are striking.
We let Claude run a small shop in the Anthropic office. Here's what happened.
Claude can now discover, learn, and execute tools dynamically to enable agents that take action in the real world. Here’s how.
Pay per crawl is a new feature to allow content creators to charge AI crawlers for access to their content.
Research on the ability of large language models to introspect on their own internal states.
Practical guidance from Anthropic on building reliable AI agents, from simple composable patterns to fully autonomous systems.
Claude Opus 4 and 4.1 can now end a rare subset of conversations. An update on our exploratory research on model welfare.
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.
New research on simulated blackmail, industrial espionage, and other misaligned behaviors in LLMs.
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…
Recent generations of frontier language models have introduced Large Reasoning Models (LRMs) that generate detailed thinking processes…
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%.
My smartest friends have bananas arguments about LLM coding.
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.
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…
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 ...
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.
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…
Research from Anthropic on the faithfulness of AI models' chain-of-thought reasoning.
We describe an approach to tracing the “step-by-step” computation involved when a model responds to a single prompt.
Building a family of AI products for everyone.
