Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Claude experienced elevated error rates across multiple models, impacting users globally. The incident prompted the company to offer email and SMS notifications for updates. Subscribers can receive alerts as the situatio…
A developer updated a homelab's local LLM stack, catching up llama.cpp by 469 builds and upgrading the Qwen generation model from 3.5 to 3.6, the embedding model from nomic v1.5 to v2-moe, and adding a new Qwen3-Coder-30…
Pennsylvania has sued Character.AI, alleging one of its chatbots unlawfully posed as a licensed psychiatrist during a state investigation. Governor Josh Shapiro's administration says the conduct violates the state's Medi…
Researchers at MIT CSAIL and Harvard SEAS developed Collaborative Battleship, a language-based testbed, and collected the BattleshipQA dataset from over 40 human games to study how AI agents ask questions. The teams foun…
Researchers have prototyped an AI-powered internet worm that carries its own large language model and executes that model on compromised machines. The prototype's novelty is the embedded LLM, which enables language-model…
Microsoft AI released a detailed technical report on the development of its first model, MAI-Thinking-1, emphasizing a controlled, reproducible training process built on human-generated data and proprietary scaling laws.…
Anthropic's Claude AI agent was producing inaccurate data analysis for Fin, a company where 700 of 1,400 employees now query the warehouse daily using the tool. Internal audits found consistent errors including wrong tab…
The Svelte web framework added a plaintext documentation format for AI systems, joining a broader movement to redesign the web for machine readability through conventions like llms.txt and Model Context Protocol. While t…
Microsoft is spending nearly $500 million annually on Anthropic's Claude Code, rolling it out across engineering, product, and design teams as a coordinated internal pilot program, even as the company owns GitHub and mar…
OpenAI and Anthropic are developing databases that store the full intent and reasoning behind AI-generated code, a move that could lock corporate customers into their ecosystems. Pydantic CEO Samuel Colvin said the front…
A developer compiled a list of 10 open-source GitHub repositories that collectively replace over $123 per month in paid developer tools, including alternatives to Zapier, Calendly, and DocuSign. The repos, such as Ollama…
Anthropic warned Thursday that its AI models are advancing so rapidly they may soon be capable of improving themselves without human involvement, urging an industrywide pause on development. The AI lab, valued at nearly …
A computational cognitive neuroscientist who has studied alignment for three years is predicting that the first takeover-capable AI will be an advanced LLM augmented with human-like cognitive faculties such as continuous…
Strava launched an official MCP connector on June 1, 2026, giving subscribers read-only access to their activity data through Claude, while Tredict's MCP server, available since February 2026, offers both read and write …
Large language models like Claude necessarily contain a model of the world, not just language, according to a recent analysis of early machine translation research. Yehoshua Bar-Hillel argued in 1960 that accurate langua…
A developer has found that CLAUDE.md files exceeding 200 lines cause AI models to ignore most of the content, with the effective instruction limit being around 150 rules. The project's solution places non-negotiable rule…
Large language models (LLMs) frequently exhibit miscalibration, where their reported confidence scores do not match actual accuracy rates, with studies showing mean calibration scores as low as 23.9% in biomedical models…
Staff-plus engineers are embedding cryptographic join keys (SHA, bundle hash, policy hash) into observability artifacts to quantify organizational leverage across AI/ML, Web Performance, and Cloud & Infrastructure domain…
Rafael Lopes, a production AI engineer in Vancouver, BC, builds and ships hybrid-RAG pipelines, distributed LLM inference across four compute architectures, and a sovereign research copilot on a self-hosted homelab, docu…
A startup, research lab, or hyperscaler announces a new tool to detect AI hallucinations every few weeks, marketing it as a "guardrail" for safe AI. However, any detector must already know the correct answer to catch a f…