391
Don’t let an LLM make decisions or execute business logic
Don’t let an LLM make decisions or implement business logic: they suck at that.
392
393
Build and Host AI-Powered Apps with Claude – No Deployment Needed
Build and share AI-powered apps directly in Claude. Update: Now available on Team and Enterprise plans. (July 31, 2025) Update: Upload PDFs, images, and more into your artifact.
394
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.
395
AI is propping up the US economy
Plus how American professors are fighting back against the AI onslaught, a backlash over AI models in Vogue, and more.
396
The "confident idiot" problem: Why AI needs hard rules, not vibe checks
We cannot fix probability with more probability. We need to re-introduce assert.
397
DolphinGemma: How Google AI is helping decode dolphin communication
Dolphin researchers are using Gemma and Google Pixel phones to try to decipher how dolphins talk to one another.
398
Less is more: Recursive reasoning with tiny networks
|| Paper | Code ||
399
It's rude to show AI output to people
Feeding slop is an act of war
400
Google AI Ultra
Learn about Google AI Ultra, Google’s new AI subscription with the highest access to our most capable models and premium features.
401
Honda: 2 years of ml vs 1 month of prompting - heres what we learned
Recalls at major automakers cost hundreds of millions of dollars a year. It’s a huge issue. To mitigate it, our company created an analytics department solely focused on categorizing warranty claims into actionable problems.
402
Anubis Works
Xe Iaso's personal website.
403
ManusAI Joins Meta
Manus is joining Meta, and we’ll continue delivering our current services while accelerating product improvements to bring more powerful, reliable general AI agent capabilities to more users and businesses.
404
AI helps unravel a cause of Alzheimer’s and identify a therapeutic candidate
A new study found that a gene recently recognized as a biomarker for Alzheimer’s disease is actually a cause of it, due to its previously unknown secondary function that triggers a pathway that disrupts how cells in the brain turn genes on and off.
405
Teaching LLMs how to solid model
It turns out that LLMs can make CAD models for simple 3D mechanical parts. And, I think they’ll be extremely good at it soon.
406
Spectral Labs releases SGS-1: the first generative model for structured CAD
Spectral Labs releases SGS-1: the first generative model for structured CAD.
407
My trick for getting consistent classification from LLMs
A technique for deterministic labeling from stochastic models, with benchmarked Golang implementation.
408
Hard problems that reduce to document ranking
There are two claims I’d like to make: LLMs can be used effectively1 for listwise document ranking. Some complex problems can (surprisingly) be solved by transforming them into document ranking problems. I’ve primarily explored both of these claims in the context of using patch diffing to locate N-day…
409
Qwen3-Omni-Flash-2025-12-01：a next-generation native multimodal large model
Qwen offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
410
Cloud Run GPUs, now GA, makes running AI workloads easier for everyone
Support for GPUs in Cloud Run makes running GPU-accelerated applications simpler, faster, and more cost-effective, as well as enabling new use cases.
411
Codemaps: Understand Code, Before You Vibe It
Codemaps is meant to offer a shared understanding of a system between humans and AI, enabling your AI to teach you about the code you are looking at…
412
Gemini 3 Pro vs. 2.5 Pro in Pokemon Crystal
From 2.5 Pro's lighthouse loop to 3 Pro's "Zombie Phoenix" strategy against Red: a full breakdown of why Gemini 3 Pro is a superior long-horizon agent.
413
Continuous Thought Machines
Introducing Continuous Thought Machines: a new kind of neural network model that unfolds and uses neural dynamics as a powerful representation for thought.
414
415
I caught Google Gemini using my data and then covering it up
I asked Google Gemini a pretty basic developer question. The answer was unremarkable, apart from it mentioning in conclusion that it knows I previously used ...
416
Compiling LLMs into a MegaKernel: A path to low-latency inference
TL;DR: We developed a compiler that automatically transforms LLM inference into a single megakernel — a fused GPU kernel that performs…
417
The Policy Puppetry Attack: Novel bypass for major LLMs
HiddenLayer’s latest research uncovers a universal prompt injection bypass impacting GPT-4, Claude, Gemini, and more, exposing major LLM security gaps.
418
Tools: Code Is All You Need
The solution to agentic flows was code all along.
419
Rob Pike got spammed with an AI slop "act of kindness"
Rob Pike (that Rob Pike) is furious. Here’s a Bluesky link for if you have an account there and a link to it in my thread viewer if you don’t. …
420
ML needs a new programming language – Interview with Chris Lattner
Listen in on Jane Street’s Ron Minsky as he has conversations with engineers working on everything from clock synchronization to reliable multicast, build systems to reconfigurable hardware. Get a peek at how Jane Street approaches problems, and how those ideas relate to tech more broadly.