Y Combinator's lab gave fifty GPT agents a survival Minecraft world and one order: build a thriving civilization. They built ...

Anthropic released v2.1.316 of Claude Code (an AI-powered programming assistance tool used in terminals and editors) on ...

Yijia Liang, founder of Compound AI and a former private-software investor, argues that the reason banks and buyout firms haven't handed real ...

Single agent vs multi-agent AI: Anthropic's Applied AI team finds one well-designed agent beats multi-agent pipelines on ...

Claude Code skip permissions gets a new env var in v2.1.311. Boris Cherny says it lifts the rm -rf ~ block. What is confirmed ...

A critical flaw in Canonical's LXD and MicroCloud let an attacker log in as a trusted client without holding that client's private key. To do it, the attacker had to reach their HTTPS API over the ...

Arena CEO Anastasios Angelopoulos reports that AI misalignment, rare on a first message, appears in more than half of conversations lasting 20 ...

Fake ChatGPT and Claude desktop apps on GitHub use DLL sideloading to deploy YobaLoader, which retrieves and executes an XMRig miner.

Google has released its next-generation frontier model, Gemini 4 Carbon, with a focus on enhancing long-duration AI agents, ...

Attackers have used fake ChatGPT and Claude desktop applications hosted on GitHub to distribute a malware chain that abuses ...

OpenHands local setup in 12 steps: run Agent Canvas with Ollama and a local Qwen model in sandboxed Docker, plus a preflight ...

Anthropic, OpenAI, Cloudflare and Google ship AI agents that find, verify and patch code bugs. How they compare.