Full-text search across 70389 articles. Combine with topic and date filters; results sorted by relevance.
A study of four open-weight large language models found that refusal behavior concentrates in sparse component mechanisms comprising 28–48% of upstream attention and MLP components, which retain 88–101% of full steering …
A new arXiv paper (2610.07093v1) benchmarks lightweight open-source language models for classifying smart data models (SDMs) on resource-constrained IoT edge devices, comparing general-purpose, reasoning-specialized, and…
A neutrosophic ensemble of Random Forest, XGBoost and Logistic Regression reached 100.00 percent accuracy on three of four held-out CWRU bearing-fault loads (92.27 percent on the fourth) but collapsed to 40.64 percent on…
AegisFlow, a multi-agent framework described in arXiv paper 2610.06971v1, cut Mean Time to Repair for fragile data pipelines by 98.1 percent, from an average of 170 minutes per patch to 3.2 minutes, with a 92 percent pat…
Researchers proposed CausalBind, a causal-modeling method for protein-molecule virtual screening that formalizes binding as sparse cross-modality interactions via a V-structure causal model under Heckman-style selection,…
Researchers released CroissantMiner, the first benchmark for end-to-end evaluation of Croissant metadata extraction from ML dataset documentation, comprising 602 papers — 102 with human-validated gold annotations and 500…
Meta rushed to fix multiple KVM escape vulnerabilities in its Muse AI system in the weeks before the product's September 8 launch, including at least one tied to CVE-2026-53359, a 16-year-old Linux kernel flaw in KVM's x…
Researchers Dhruv Sahnan, David Corney, Irene Larraz, Giovanni Zagni, Ruben Miguez, Zhuohan Xie, Iryna Gurevych, Elizabeth Churchill, Tanmoy Chakraborty and Preslav Nakov published a study in Transactions of the Associat…
Bees, an open-source agentic harness released under MIT/Apache 2.0 licenses, runs agentic teams locally on macOS, Windows and Linux, reading user files in place on their own disk. The tool lets users describe work in pla…
ClawSecure's AI agent threat report found that the MCP protocol specification itself, not individual vendor code, allows attacker-controlled links to be auto-fetched by MCP servers on Linear, Notion, and Dropbox Dash, an…
Scammers are impersonating AI chatbots including ChatGPT and Gemini through a browser-in-the-browser technique that creates fake sign-in windows to capture login credentials and multi-factor authentication codes in real …
Google released EmbeddingGemma 2, a 740M-parameter multimodal embedding model that maps text, code, images, video, and audio into a single unified embedding space under an Apache 2.0 license. The model offers flexible ou…
Google announced Google Playground, an "experimental gaming platform" that lets users create, play and share fully custom games with generative AI and no coding experience, according to a Google spokesperson. Playground …
Google released EmbeddingGemma 2, an open embedding model with 740 million parameters that converts text, images, video, audio, and code into vectors and runs on-device using about 191 MB of RAM. According to Google, Emb…
Multiple developers on Hacker News and personal blogs report that OpenAI's Codex consumes 3 times more tokens than Anthropic's Claude for identical coding tasks, with one user spending $150 in credits to fix a single pro…
Texpile, an open-source Typora-style editor for LaTeX and Typst, launched on Hacker News' Show HN with live preview that renders most changes in under 20 ms by running its own script inside the user's existing lualatex i…
Termalin released a watch grid that mirrors every open SSH session as a live tile in a single view, highlighting sessions being driven by AI agents such as Claude Code, Codex, and Gemini working through its local MCP ser…
A single developer, working with an AI collaborator on a roughly $100/month subscription, produced a formal C17 compiler written entirely in ARM64 assembly that self-hosts to a byte-identical binary, cross-compiles to x8…
A developer outlined a lightweight workflow for linking architecture decision records (ADRs) to the git commits that implement them, using a date-plus-slug filename convention, a Markdown template, and git commit trailer…
NVIDIA released AICR v1.0, an open-source AI Cluster Runtime that provides version-locked, validated recipes for GPU-accelerated Kubernetes cluster configuration, establishing a stable compatibility contract across its C…