Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
The Trump administration imposed export controls on Anthropic after a 24-hour effort to convince the company to voluntarily pull its new AI model Fable, which officials believed posed security risks. The decision followe…
OrchestKit's documentation site implements a stack of standard files—llms.txt, OpenAPI specs, MCP endpoints, and .well-known identity files—to make its docs machine-readable for AI agents. The project, a free MIT-license…
Zerostack released v1.5 of its lightweight Unix-inspired coding agent, achieving feature parity with Claude Code on key agent capabilities. The update introduces a redesigned TUI, multi-modal input support, advisors for …
Apple's iOS 27 developer beta includes a dormant Extensions framework that would let users select third-party AI providers like ChatGPT, Claude, and Google Gemini within Siri, but the feature is disabled on Apple's backe…
A developer built a Claude Code plugin marketplace as a single GitHub repo with a specific folder structure, enabling versioned plugin management across machines. The marketplace uses a `marketplace.json` registry and su…
A developer proposes replacing fragile CSS selectors with LLM-powered zero-shot JSON extraction for web scraping. By converting cleaned HTML to Markdown and using a predefined JSON schema, the approach makes pipelines re…
Sophos X-Ops analysts detected a modular post-exploitation framework that used AI-assisted tooling and automated workflows to test endpoint detection and response (EDR) evasion, according to a Sophos blog post. The frame…
A developer released a local .NET multi-agent LLM pipeline AI therapist on GitHub, using a compact wire format to enable offline therapy simulations. The project aims to provide accessible mental health support without r…
WastedCode released TrueCast, a CLI and plugin system that installs portable expert personas—such as product manager, architect, and security reviewer—into Claude Code sessions. The tool provides versioned, sandboxed sub…
Apple's AI crisis is easing, according to Bloomberg's Mark Gurman, as the company prepares a major Siri redesign and AI capabilities for iOS 27, with an internal deadline of spring 2026. The turnaround is expected to be …
A developer maintaining bulwark-mcp, an open-source proxy that scans MCP tool results for indirect prompt injection, discovered that a class of attacks bypasses the detector by impersonating the security pipeline itself.…
A PhD student released CMG, an open-source tool that audits LLM-based judges by requiring them to back each verdict with explicit claims tied to evidence, flagging untrustworthy decisions without using a second model. Th…
Mlx-optiq, a new open-source tool, enables per-layer mixed-precision quantization of large language models on Apple Silicon, allowing users to run, fine-tune, and serve LLMs locally on Macs without a GPU cluster. The too…
A backend engineer at a major tech company spent three months struggling with AI coding tools that failed to retain project context, forgetting decisions and reasoning overnight. The engineer built a self-filtering memor…
Cohere launched North Mini Code, its first developer-focused coding model, targeting developers who want controllable AI that runs on their own infrastructure. The release extends Cohere's enterprise strategy into the de…
Apple built a third-party AI Extensions system for Siri in iOS 27 beta, allowing users to choose between ChatGPT, Claude, and Gemini, but skipped announcing it at WWDC due to EU regulatory standoff, an OpenAI lawsuit thr…
Mantic Think launched a free, private web UI for Ollama models that allows users to chat with Ollama Cloud models using their own API key or connect to a local Ollama instance. The platform requires no account or signup,…
A Hacker News user shared gmd, an open-source tool that indexes local markdown files with full-text, vector, and hybrid search via Typesense, and includes web search, fetch, crawl, research, LLM-wiki patterns, and agents…
A developer building an agent-based project describes a pragmatic approach to guardrails: each check in the toolchain was added only after a real failure occurred, not designed upfront. The method, illustrated by a gloss…
The US government's June 12 order barring foreign nationals from accessing Anthropic's Claude Fable 5 and Mythos 5 models led to a global takedown of both models, exposing a compliance gap in frontier model APIs. The res…