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 bank's AI loan assistant triggered a $1.2 million regulatory fine within 48 hours after misquoting a compliance clause, exposing the inadequacy of single "hallucination metrics" like BLEU or ROUGE for legal risk. A six…
A developer found that organizing a filesystem for AI agents rather than humans significantly improves efficiency, as Claude Code automatically loads `CLAUDE.md` files from the directory tree based on the agent's current…
Alibaba's Qwen team released Qwen3.7-Plus, a multimodal agent model that integrates visual perception, GUI operation, and coding into a single autonomous loop. In a demonstration, an agent built on the model independentl…
A developer released a tool that lets users paste a YouTube video URL into an AI chat to receive a text summary of the video's content. The tool aims to save time by eliminating the need to watch full-length videos for k…
A Hacker News user asked the community what they would name their own large language model, prompting responses that compared the AI to an overeager PhD student—brilliant and hardworking but lacking real-world experience…
A developer has outlined how to force AI coding assistants to generate modern Java 26 code by adding strict, opinionated workspace rules to `.cursorrules` or `.claudecode` configuration files. The rules ban legacy Java 8…
A developer detailed a fully open-source AI stack that runs in production with zero API costs, addressing the financial reality of scaling AI applications where every new user adds a fixed marginal cost. The stack uses O…
A developer created MemBot AI, a customer support assistant with persistent memory that stores and retrieves customer issues, preferences, and conversation history across interactions. The system maintains context from p…
Leonie Monigatti presented a workshop titled "Agentic Search for Context Engineering" at AI Engineer Europe 2026 in London on April 8, 2026, arguing that context engineering is 80% agentic search. The workshop traced the…
TwitterAPI.io has published a decision guide for developers choosing between its REST API and hosted MCP server at `https://mcp.twitterapi.io/mcp`. The REST API, with 70 endpoints, is recommended for high-throughput pipe…
Tencent appointed former OpenAI researcher Yao Shunyu, 27, as Chief AI Scientist in December 2025 to lead the company's push toward artificial general intelligence. The Chinese tech giant will double its AI spending to o…
A developer discovered a logic flaw in AI-generated authentication middleware that would allow an attacker to maintain a session indefinitely after obtaining a single valid refresh token. The vulnerability, found in prod…
Jürgen Schmidhuber, a pioneer in artificial intelligence whose lab developed foundational ideas like LSTM, world models, and artificial curiosity decades before they became mainstream, argued in a new interview that larg…
A developer released Nanocode-CLI, a lightweight terminal-based AI coding assistant written in Python that operates entirely within a command-line interface. The tool features live turn control, file-state tracking, stal…
A 2-billion-parameter fully continuous, end-to-end autoregressive text-to-speech system called dots.tts has achieved state-of-the-art performance across multiple benchmarks, including the best average results on Seed-TTS…
A developer who uses AI tools daily has compiled a scorecard comparing AI and human capabilities, finding that AI wins on speed, scale, and pattern-matching while humans retain advantages in judgment, accountability, and…
Anthropic's Claude Code can be redirected to any compatible backend using a single environment variable, `ANTHROPIC_BASE_URL`, enabling caching, rate-limit pooling, multi-vendor routing, and audit logging without code ch…
University of Toronto researchers demonstrated a proof-of-concept adaptive AI worm that autonomously identifies and exploits known vulnerabilities across devices, achieving a 73.8% compromise rate in a simulated 33-machi…
Vercel has released a guide for developers to build AI-powered Slack agents using Chat SDK and AI SDK, enabling autonomous responses to mentions, conversation history tracking, and tool calling without managing infrastru…
HelperX, an AI-powered X automation tool, has generated over 100,000 automated replies that avoid generic responses by using a layered prompt structure and persona-based customization. The system employs five controllabl…