Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
An AI security company faces a lawsuit for publishing hallucinated vulnerability findings, highlighting a crisis of trust in AI-generated security research. The case follows cURL's shutdown of its bug bounty program due …
Elon Musk is reassigning dozens of SpaceX's top engineers, including those from recently acquired Cursor, to overhaul the Grok AI model, aiming to challenge what he calls the 'woke monopoly' of OpenAI and Google. The mov…
Meta AI's Brain2Qwerty system decodes brain signals into text using non-invasive MEG sensors, achieving a 32% character error rate, but the technology is not yet real-time or commercially viable. The research demonstrate…
A developer created cmux-orchestrate, a Claude Code skill that enables driving cmux terminal panes, splits, tabs, and agents via the cmux CLI. The skill provides comprehensive control over terminal layouts, workspaces, w…
A developer created 'orchestrate-plan', a Claude Code skill that executes a locked PLAN.md by delegating implementation to cmux worker panes one phase at a time, clearing context between phases. The skill supports parall…
Cortex's 2026 benchmark reports a 30% industry-wide rise in change failure rate (CFR) since AI coding adoption accelerated. A developer provides a practical guide to measuring CFR using DORA's formula and existing data, …
A developer building an LLM-based email classifier identified a critical design flaw: the model's self-reported confidence score is used as a gate for autonomous actions, but a confident impersonation could bypass it. Th…
JPMorgan analyst Harlan Sur reaffirmed that Broadcom's TPU v9 chip for Google remains on schedule for 2028 production, raising Broadcom's price target to $580 amid recent delay fears. The 2nm ASIC program, featuring four…
A developer introduced CAPE (Collaborative Agents Prompt Engineering), a multi-agent framework that models a software product team with four specialized roles: Product Owner, Designer, Developer, and Cape Master. The fra…
Developer Maneshwar is building git-lrc, a free and source-available Micro AI code reviewer that runs on every commit. The project highlights the critical need for data privacy in AI systems, warning that models can memo…
Stephen Gruppetta explains that AI agents differ from plain LLMs by combining a model with instructions, memory, tools, and an execution loop. He clarifies that an LLM's context window is not true memory, as it only carr…
The Postman plugin for Claude Code reduced its largest skill's token cost by 60% and overall session overhead by 65% through progressive disclosure and removal of redundant skills, saving roughly 3,600 tokens per session…
Mistral launched Vibe, a unified AI agent that replaces Le Chat and combines work assistant and coding agent capabilities under a single license. The agent runs on Mistral's own models and infrastructure, targeting enter…
AI coding platforms are shifting from unlimited plans to token-, credit-, and quota-based subscriptions. Five plans—MiniMax Token Plan, MiMo Token Plan, GLM Coding Plan, and others—offer developers flexible usage at comp…
Anthropic introduced the Model Context Protocol (MCP), an open standard that standardizes how AI applications connect to external tools and data sources, replacing custom integrations with a shared protocol. MCP uses a c…
Developer Simon Willison built agent-historian, a tool that lets AI coding agents query their own session history, after repeatedly encountering agents that forget solutions from previous sessions. The tool treats past t…
Cheap AI intelligence is repricing labor, shifting work from human execution to machine delegation. Five forecasts predict that prompts become work orders, agents replace chatbots, and companies treat tokens as labor sub…
Coinbase CEO Brian Armstrong outlined five AI cost controls that cut the company's AI bill by nearly half while allowing engineers to increase token usage, including defaulting to cheaper open-weight Chinese models, rout…
A non-technical user proposes creating an ultra-simple one-click tool for running open-source LLMs locally, citing installation complexity, hardware compatibility issues, and inaccessible interfaces as key pain points. T…
Google capped Meta's use of its Gemini AI model after Meta exceeded its computing capacity, according to sources. The incident highlights that even major tech companies face computing power shortages despite massive data…