Google adds Gemini to Chrome across seven APAC countries; Moonshot AI’s open-weight Kimi K2.6 runs up to 300 coordinated agents for long-horizon coding; Chinese workers urged to train “Colleague Skill” clones; OpenAI rolls out GPT-5.4-Cyber for vetted defenders.

Quick roundup: Google’s A2UI lets agents build native UIs on the fly, Ray-Ban Meta + OpenClaw explores always‑perceiving wearable AI, Gemini’s Personal Intelligence trades privacy for context, and NVIDIA’s Ising opens hybrid quantum-classical AI research.

Roblox's AI agents can plan, build and test games from scratch. Google ships a native Gemini Mac app that can read your screen for context. Luma launches an AI-powered studio starring Ben Kingsley in a Moses series, while Bytedance unveils Seedance 2.0 but geoblocks the US amid copyright and legal tensions.

OpenAI expanded its Agents SDK to help enterprises build safer autonomous workflows. DeepMind’s Gemini Robotics-ER 1.6 adds visual and spatial reasoning to robots. An open-source project went proprietary after AI-assisted hacks. MIT warns some tasks may be "minimally sufficient" to AI by 2029—upskill now.

Quick AI roundup: Chrome adds AI Skills to save Gemini workflows; Nvidia and UMD release Audio Flamingo Next, an open audio-language model that reasons over complex sounds and music; MIT explores how AI is reshaping software engineering; Google Home update fixes Gemini interruptions.

Today’s AI shakeup: global GPU shortages and outages, OpenAI shelving Sora while prepping a new model ‘Spud’, a leaked memo accusing Anthropic of inflated revenue, the Hiro acquisition to add finance tools, and Vercel benefiting from an AI hosting boom. What the compute crunch means for the AI race.

From a stalking victim suing OpenAI over ChatGPT-fueled delusions to a study showing agentic skills fail in real-world tests, plus Meta & KAUST's 'Neural Computers' concept and MiniMax open-sourcing M2.7 — a self-evolving model now on Hugging Face. Big legal, technical, and open-source shifts in AI.

ChatGPT unveils a $100 Pro tier for heavy users, while a landmark Florida investigation probes OpenAI's alleged role in a deadly shooting. Stanford research questions whether multi-agent gains come from extra compute or true collaboration, and Google Gemini adds interactive in-chat data visualizations.

A rapid AI news roundup: Poke aims to make AI agents as easy as texting, Stability AI unveils Brand Studio for brand-consistent imagery, a Muckrack study finds 25% of chatbot citations trace to journalism, and Databricks’ Matei Zaharia provocatively claims AGI is already here.

A 26-person startup builds a high-performing open-source LLM, Meta plans partial open-sourcing of its models, why an “agent-first” redesign could upend enterprise workflows, and Google Maps is auto-generating photo captions with Gemini — quick takes on where AI is headed.