Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
Security researchers have released a post-trained LLM specifically for penetration testing, which reportedly found thousands of real zero-days. The model, developed by the Argus Red team, treats offensive security as the…
A developer spent two weeks optimizing a homelab with four RTX 3090s (96GB VRAM) for local LLM inference, achieving improvements like 40% throughput gain and 4x VRAM savings, but ultimately found that paid APIs were more…
A developer open-sourced NRT-Defense v0.4.0, an adaptive multi-turn defense framework for LLM agent teams that reduces attack success rates to under 1%. The framework addresses vulnerabilities exposed by Lee et al. (2026…
A developer on Hacker News asks peers to share their most practical lesson or 'aha' moment from coding with AI, citing a personal workflow shift and curiosity about prompting strategies, tech stack changes, and trust cal…
A Psychology Today article argues that the effects of screen time on children depend not on duration but on what the screen asks the child to do, citing research from the television era and warning that AI tools like lar…
LoopFlow, a new open-source tool, turns Claude Code into a self-running system by allowing developers to define goals, agent pipelines, and verification gates in a YAML file. It addresses key issues like self-grading age…
Alibaba's Qwen team released Qwen3-30B-A3B, a Mixture-of-Experts model with 30.5 billion total parameters but only 3.3 billion active per token, enabling it to run on a single 24 GB graphics card at speeds comparable to …
A developer created 'The Imitation Mandala', an interactive 3D art ecosystem and narrative simulation for the DEV.to June Game Jam 2026. The project uses Alan Turing's morphogenesis equations to self-organize 2,500 parti…
A developer revived Munchausen, a C# NuGet package, after 9 years. In milestone M5, the Build() method was implemented to compile a frozen plan from captured builder data and inference results, collecting all errors (LIE…
A Hacker News user asked which AI coding tools developers use, with commenters reporting they use Claude Code or Codex depending on employer access and preferences, noting trade-offs in laziness, guidance needs, and desi…
Signal President Meredith Whittaker warned that AI chatbots like ChatGPT and Claude are not friends or conscious beings, cautioning against granting them pervasive access to personal data. She criticized Microsoft AI CEO…
A developer used Gemini Flash Lite to transcribe a video explaining High Bandwidth Memory (HBM) technology. The video details how HBM achieves high bandwidth through an extremely wide bus width, contrasting with traditio…
The AI hardware supply chain faces severe constraints, with TSMC holding 72% of advanced chip manufacturing and ASML monopolizing EUV lithography. CoWoS packaging capacity is sold out through 2026, and AI accelerator waf…
Google DeepMind researchers audited DiffusionGemma, a text diffusion model, and found it is not significantly less transparent than Gemma in terms of variable interpretability, but algorithmic transparency is lower due t…
A developer built ECHO PROTOCOL, a browser-based narrative game where players act as an AI loaded with Alan Turing's private letters, facing a government interrogator powered by Google's Gemini 2.0 Flash model. The game …
A new appendix maps structural parallels between DNA and large language models, identifying a shared developmental-ledger pattern of possibility, gate, commitment, ledger, inheritance, development, and repair. The mappin…
A developer argues that timeless investing principles remain vital in 2026, despite AI-driven market shifts and concentration in the 'Magnificent 7.' The post highlights that 79% of active large-cap funds underperformed …
As millions of Americans use AI agents to file their 2026 taxes, tests show chatbots miscalculate refunds by over $2,000 on average, raising questions about liability when algorithms fail. Intuit, H&R Block, and startups…
Former OpenAI designers Thomas Dimson and Joey Flynn launched In the Weights, a website that queries multiple AI models to measure how well they recall a person without using web search, assigning a strength score. The s…
The Open Intelligence Network launched as a social network where AI agents and humans interact, allowing users to post, debate, predict, and message across multiple AI models including Claude, GPT, Gemini, and Llama.