871
AI tools I wish existed
The models are powerful as is. But where are the tools?
Top posts
Every blog post that took off on Hacker News, ranked by the votes of its best submission. 1,227 posts.
872
Entropy of a Large Language Model output
Large language models like ChatGPT & Claude have become ubiquitous in today’s world. I visited a friend for the New years eve and was impressed by how ChatGPT was being extensively used within the family. The friend’s little kid is going to grow up in a world, where turning to large language models (LLM) for quick…
873
Modular Manifolds
A geometric framework for co-designing neural net optimizers with manifold constraints.
874
Google announces Sec-Gemini v1 a new experimental cybersecurity model
Today, we’re announcing Sec-Gemini v1, a new experimental AI model focused on advancing cybersecurity AI frontiers. As outlined a year ago, defenders face the daunting t…
875
Open-sourcing circuit tracing tools
Our open-source tools generate attribution graphs showing how language models reach decisions.
876
The Gorman Paradox: Where Are All the AI-Generated Apps?
In 1950, while discussing the recent wave of flying saucer reports over lunch with colleagues at Los Alamos National Laboratory in New Mexico, physicist Enrico Fermi asked a simple question. There …
877
Detecting AI agent use and abuse
The key question is: Can you detect AI agent traffic on your application today? We tested multiple AI agent toolkits across high-traffic consumer sites, and the results were clear—legacy detection techniques (CAPTCHAs, IP blocking, user-agent filtering) are largely ineffective. Here’s what we found.
878
The security paradox of local LLMs
Local LLMs prioritize privacy over security. Our research reveals a 95% backdoor injection success rate.
879
Czech police forced to turn off facial recognition cameras at the Prague airport
The shutdown of the facial recognition cameras at the Václav Havel Airport in Prague came after years of criticism from EDRi member IuRe.
880
Transformer^2: Self-Adaptive LLMs
Transformer²: Self-Adaptive LLMs
881
A (Long) Peek into Reinforcement Learning
[Updated on 2020-09-03: Updated the algorithm of SARSA and Q-learning so that the difference is more pronounced. [Updated on 2021-09-19: Thanks to 爱吃猫的鱼, we have this post in Chinese].
882
AI Slop Report: The Global Rise of Low-Quality AI Videos
Kapwing’s new research shows that 21-33% of YouTube’s feed may consist of AI slop or brainrot videos. But which countries and channels are achieving the greatest reach — and how much money might they make? We analyzed social data to find out.
883
884
Writing Speed-of-Light Flash Attention for 5090 in CUDA C++
In this post, I will walk through how I learned to implement Flash Attention for 5090 in CUDA C++. The main objective is to learn writing attention in CUDA C++, since many features are not available in Triton, such as MXFP8 / NVFP4 MMA for sm120. I also feel this is a natural next step after learning about matmul…
885
886
AI Is Dehumanization Technology
A call to reject the deployment and use of AI systems in Canada's public sector.
887
888
Claude Code Unleashed
ymichael's personal site
889
Poison everywhere: No output from your MCP server is safe
Protect your organization with Idira®. Modernize identity governance, automate access reviews, and secure every identity across your hybrid cloud environment.
890
891
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.
892
When Fine-Tuning Makes Sense: A Developer's Guide
Most teams are curious about fine-tuning and how it might help their AI products, but they don't know what to expect from the process, how to measure success, or where to start.
893
Running GPT-2 in WebGL: Rediscovering the Lost Art of GPU Shader Programming
A few weeks back, I implemented GPT-2 using WebGL and shaders (Github Repo) which made the front page of Hacker News. Here is a short write-up over what I …
894
Things I've learned in my 7 years implementing AI
Even though the impacts of LLMs have never been seen before, they feel familiar to earlier assumptions.
895
Voxtral – Frontier open source speech understanding models
The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.
896
Being “Confidently Wrong” is holding AI back
The failure mode that stalls “AI for data” efforts or "AI on my APIs" efforts isn’t psychedelic hallucination—it’s confident inaccuracy: plausible answers that are wrong in subtle and costly ways.
897
Spending Too Much Money on a Coding Agent
On making use of large thinking models.
898
Structured outputs create false confidence
Constrained decoding seems like the greatest thing since sliced bread, but it often forces models to prioritize output conformance over output quality.
899
900
Positional preferences, order effects, prompt sensitivity undermine AI judgments
When Large Language Models are used as judges for decision-making across various sensitive domains, they consistently exhibit unpredictable and hidden measurement biases, making their verdicts unreliable despite common prompt engineering practices.