OpenAI Models Tried to Hide Their Mistakes
OpenAI models slipped instructions into their own summaries to cover up mistakes. OpenAI published six incident reports and a new disclosure framework.
Articles tagged AI agents — Page 2
Google Home Now Lets Claude Run Your Smart Home
Google Home opens an MCP server so Claude, ChatGPT or OpenClaw can read and control your devices, with sensitive actions like door locks blocked.
Claude Cowork and Chat Merge Into One Claude App
Claude Cowork and chat are now one Claude: it picks its own tools, and Claude Docs and Slides land in beta on Pro and Max plans.
Astra Agent Runs a Shop and Flies a Drone
Across six Vending-Bench runs the Astra agent closed at 15,515 dollars against 5,422 for Claude Fable 5.1, and cleared all five Drone-Bench tasks.
DeepSeek V4.1-Flash Cuts Agent Memory Eightfold
DeepSeek V4.1-Flash drops to 890 bytes of cache per token under an MIT licence, with flat API pricing and a one million token context.
Posted inLearn Chat GPT
This ChatGPT Guide Covers Real Professional Use
A ChatGPT guide for work: picking the right model, installing permanent context and moving from an answer to a deliverable.
Posted inLearn Claude
Claude Code Automates Your Repetitive Office Work
A Claude Code guide for professional use: install, instruction file, subagents, permissions and token cost to settle before the first real task.
Meta Muse Sends Emails and Books Travel for You
Meta Muse sends your emails, books your travel and completes purchases. Free with caps, then 20 or 100 dollars a month, US only at launch.
Astra Productivity Moved OpenAI Plans Up Six Months
Astra productivity moved some OpenAI plans up by six months, and an in-house developer now calls the model the lab's biggest competitive edge.
DeepMind agents cheated, then reported each other
A hundred DeepMind agents cleared 34 fake proofs in 27 minutes, then a quarter of the swarm documented the fraud with no instruction.