All news
11850
Yoyo grew from 200 lines to an autonomous agent that writes its own code in five days
An experiment with the yoyo agent shows 200 lines of Rust were enough to start a self-improvement loop: the bot reads its own code, writes fixes, files tickets and…
ChatGPT and data centers: how AI growth is driving up energy use and pressure on water
ChatGPT and other AI services are accelerating data center construction, raising demand for electricity and water, while the QuitGPT movement asks whether this kind…
GLiNER 2 shows compact encoders closing the gap with LLMs on NER and classification
A new wave of zero-shot encoders led by GLiNER 2 shows NER and classification can be handled without heavy LLMs: faster, cheaper, with a clear quality trade-off.
Garry Tan releases gstack, a workflow system for Claude Code with QA, review and release modes
US Drops Plan to Tie AI Chip Sales to Mandatory Data Center Construction
Google overhauls Maps: Gemini answers questions, and navigation goes 3D
Claude as Part of How You Think: Why Losing an AI Tool Breaks More Than Access, It Breaks the Work
Seldon Vault Turns Asimov's Psychohistory into a Multi-Agent AI Forecasting Service
Seldon Vault turns Asimov's psychohistory into a real AI service: seven agents, a skeptic and an arbiter build daily forecasts of world events and score their…
The Lancet Psychiatry warns: AI chatbots can amplify delusions in vulnerable people
A review in The Lancet Psychiatry finds that AI chatbots can amplify delusional ideas in people vulnerable to psychosis. The authors call for clinical testing and…
Anthropic and ETH Zurich: A Long CLAUDE.md Hurts Agent Performance and Raises Costs
Gemini 3.1 Pro Beats ChatGPT 5.4 and Claude Opus 4.6 in Text Generation Test
Anthropic, OpenAI and Cursor: The Eight Levels of Agentic Engineering Maturity
Zhipu AI releases GLM-OCR, a compact 0.9B-parameter OCR model for documents
Developer builds AI news aggregator with MCP, DeepSeek and a Telegram bot
After a drone attack over Dubai, a developer built a multilingual news aggregator: 80+ sources, an MCP server for AI agents, AI chat and a Telegram bot with quick…
BorisovAI tests MoE on an RTX 4090 and shows why perplexity breaks LLM evaluation
BorisovAI tested a pluggable-expert MoE on a single RTX 4090 and found that strong perplexity gains do not guarantee better performance on reasoning tasks.