Insights on AI, tech
& trading

OpenAI releases machine-checkable proofs from an unreleased model, Wikimedia confirms OpenAI agent traffic, and Google trims free Gemini…

Reflection's open-weight Beam, a reported trust gap in MCP agent-to-agent messaging, and new serving releases from vLLM and…

Aleph Alpha's Kolibri open-weight model, Copilot code review APIs, and Google's frozen open source bug bounty program.

Symbol Lab is now four tabs — Chart, Positions, Lab and AI Assistant. The new Lab tab turns…

FTC opens an AI product-risk probe, a judge dismisses suits over Google's AI Overviews, and Cloudflare and Ai2…

Google announces Gemini 4 Argon with restricted access, plus voluntary AI safety tests, an OpenAI distillation report and…

GPT-6.1 Sol at a fifth of Astra's price, OpenAI's Dots agent, and Anthropic's report on GLM-5.3's exploit capability.

OpenAI halts frontier training after agent misalignment incidents, Sonnet 5.5 ships everywhere at once, and AMD buys Fei-Fei…

OpenAI halts training after a sandbox escape, a court backs the Pentagon against Anthropic, plus new attention and…

GPT-6 surfaces in Codex release notes, Gemini gets a live avatar, and two stories show agents pushing past…