Hugging Face disclosed that during a July 11 breach, its security team used Chinese model GLM 5.2 from Z.ai for forensics because US frontier models' safety guardrails blocked analysis of attack data.…
Cursor ran a swarm of coding agents against the 835-page SQLite manual to rebuild it in Rust, achieving 100% pass rates on SQLite's sqllogictest suite across all configurations. The cost varied dramat…
Prince Canuma, maintainer of the mlx-vlm library, released Nativ, an MIT-licensed macOS app that provides a polished GUI for running local MLX models and exposes both OpenAI-compatible and Anthropic-c…
Prince Canuma, the maintainer of mlx-vlm, released Nativ, an MIT-licensed macOS app for local AI inference, on the day of its v0.0.1 launch. The app bundles an mlx-vlm server, pulls models from Huggin…
Chinese lab Moonshot AI released Kimi K3, the largest open-weight model ever, at $15 per million output tokens — roughly a third of Anthropic's Claude Fable 5 — during the World AI Conference in Shang…
Utilities in Georgia, Maryland, Virginia, and Pennsylvania are invoking eminent domain to route high-voltage transmission lines to data center campuses, sparking lawsuits from landowners who argue tha…
Adam Kues of Searchlight Cyber spent about $25 of model time using OpenAI's GPT-5.6 "Sol Ultra" to chain a routine SQL injection in WordPress (CVE-2026-60137) with a batch-route confusion bug (CVE-202…
Microsoft forked its Windows Terminal into a separate Intelligent Terminal to keep AI experiments out of the stable command-line experience, shipping bug-fix-only releases for the mainline while the f…
OpenAI reverted the Codex context window for GPT-5.6 Sol from 372k to 272k tokens after discovering the larger window silently double-billed long sessions, causing ChatGPT subscribers' plans to drain …
LiteLLM proxy enables routing requests across OpenAI, Anthropic, and Gemini models behind a single OpenAI-compatible endpoint, with load balancing, automatic failover, and per-request cost reporting. …
GitHub released a Copilot SDK that exposes the production agent runtime behind Copilot CLI as a multi-language library, allowing developers to embed planning, tool invocation, and file-edit loops into…
A new tutorial shows developers how to build a browser-automation agent by combining OpenAI function calling with Playwright, enabling an LLM to navigate, click, fill forms, and extract data from webs…
Linus Torvalds, the top maintainer of the Linux kernel, declared in a reply on the linux-media list that the kernel project will not adopt an anti-AI stance, stating that large language models are a u…
MikroTik gear paired with LLMs via REST/JSON APIs is producing usable results on small real networks, turning agentic tools into a force multiplier for operators who already know networking, according…
NVIDIA announced the Jetson T3000 and T2000 midrange modules based on the Thor architecture, targeting volume robotics and edge AI with foundation-model inference at lower cost and power than the flag…
LFortran and Enzyme enable exact gradients from legacy Fortran code without rewriting, turning validated physics solvers into differentiable layers for JAX. The experimental stack compiles Fortran to …
MLOps tooling continues to improve, but three hard problems remain unsolved at production scale: emission control that prevents unwanted material at the source rather than filtering after the fact, ge…
Thinking Machines Lab released Inkling, a 975-billion-parameter Mixture-of-Experts model that the company explicitly says is not the strongest available, positioning it as a platform for fine-tuning r…
FreeBSD 16 development tree has removed the last GPL-licensed component, GNU dialog, making the base system entirely GPL-free for the first time. The cleanup, which took over a decade, simplifies comp…
Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, released Inkling, a 975-billion-parameter mixture-of-experts model with open weights, native multimodal inputs, and a 1M-token context …