Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A new benchmark, IatroBench, reveals that heavily safety-trained AI models cause iatrogenic harm by withholding critical medical information from patients while providing it to physicians, based on 3,600 responses across…
TakoQA, a new open-source testing tool, uses a swarm of LLM-driven browser agents to autonomously break web applications by executing plain-language missions. The tool observes pages visually, decides actions via an LLM,…
A developer shares a workflow for turning Claude-generated HTML artifacts into live preview URLs. The approach covers single-file uploads, CLI deployment, and integration with agentic coding tools, emphasizing that only …
A developer introduced SQLazy, a tool that compiles natural-language step-by-step logic into auditable SQL, addressing the trust crisis in AI-generated SQL. Research shows LLMs achieve only 64.5% accuracy on Text2SQL tas…
A developer turned their gaming PC into a local AI homelab to avoid cloud token costs and rate limits. By running a Vision Language Model (VLM) locally, they can extract text from screenshots in about 8 seconds without s…
A consensus is emerging on securing the agentic software development lifecycle (SDLC) as AI coding agents increasingly ship to production, but ready-to-use security solutions remain elusive. The number of pull requests h…
A tech commentator argues that keyboard and mouse interfaces are unnatural for humans and predicts their obsolescence within 20 years, citing innovations like voice models, Neuralink, and Apple Vision as evidence of a sh…
Anthropic accused Alibaba of running 28.8 million API queries through nearly 25,000 fraudulent accounts over 45 days to distill Claude's software engineering and agentic reasoning capabilities. The attack, disclosed on J…
Alibaba's Wan Streamer v0.1, an end-to-end real-time interactive foundation model, achieves sub-second audio-visual response latency by processing language, audio, and video in a single Transformer with block-causal atte…
Open-source coding agents have matured to rival Claude Code in 2026, offering model-agnostic, auditable, and self-hostable alternatives. Leading projects include OpenCode (161k GitHub stars), OpenAI Codex CLI (74k stars)…
Prompt engineering is not disappearing but evolving into context engineering, argues a developer who builds AI workflows. As AI models improve, the need for clear, structured prompts grows, especially in production syste…
OpenCode, a 178,218-star open-source AI coding agent built in TypeScript, offers hidden capabilities beyond basic autocomplete, including support for 30+ LLM providers, a plugin system, and MCP OAuth flows. The tool allo…
DeepSeek V4's pricing is nearly 50x cheaper than Anthropic and OpenAI's frontier models, raising concerns that the US companies may rely on fearmongering and government bans on open-weight models to maintain high prices …
A developer with over a decade of web development experience describes their journey using AI tools like Claude Code, emphasizing the importance of harness engineering and critical oversight. They caution against accepta…
A former software company founder reports that at a friend's 15-person firm, developers rely heavily on AI tools like Claude, with code no longer serving as the source of truth, code reviews done by AI, and multiple simu…
Anthropic's negotiations with the Trump administration to lift export controls on its Claude Fable 5 AI model have improved after CEO Dario Amodei was replaced by co-founder Tom Brown in talks, according to a Wired repor…
Google delays the launch of its Gemini 3.5 Pro AI model from June to July 2025, citing feedback from early testers and tuning for agentic tasks. The delay coincides with the departure of key researchers Noam Shazeer to O…
Tom MacWright argues that AI-generated resumes, portfolios, and outreach emails strip away the human connection and personal character that hiring managers and collaborators rely on. Drawing on 12 years of reviewing job …
A developer documented workarounds for breaking changes in Anthropic's Claude Code CLI tool, including pinning version 1.0.17, disabling auto-updates, and patching the `exit_plan_mode` popup to prevent focus disruption. …
Astromech, a startup founded in 2025 by Ben Lamm and George Church, develops a Large Life Model and autonomous biological intelligence system that fuses evolutionary logic with deep learning to decode regulatory patterns…