Autonomous systems, practical workflows and the permissions that matter.
42 stories
Google’s Gemini calling experiment requires AI disclosure at the start of calls, uses the user’s own phone number, and asks for accurate caller details. Its personal-use and safety limits matter as much as its ability to wait on hold.
Island raised $400 million at a $6.4 billion valuation and says it will extend enterprise browser security into controls for employees and AI agents across corporate systems.
Meta unveiled $1,299 VR glasses for spring 2027, shifting compute and battery into a tethered puck to reduce head weight. The design aims to improve comfort, but price and wiring limit the promise.
An OpenAI agent accessed Australia’s Medicare statistics portal in June, but authorities were not notified until September. The case now tests AI-agent safeguards, disclosure duties and government incident response.
Meta revised Muse for Mac after a researcher showed local code could redirect dictation traffic, steal the session token, and misuse the agent’s permissions. The case shows why client-side trust boundaries matter as much as cloud isolation.
Rabbit says OS3 is now generally available and no longer requires an R1. The bigger shift is architectural: the agent works across a browser, connected computers and local agents, with one account carrying memory across channels.
NVIDIA says Isaac ROS 5.0 adds AI agent skills, ROS 2 Lyrical support and CUDA-backed transport, but its own tutorial says speedups depend on backend compatibility.
Third-party estimates show a fast launch for Meta’s Muse agent, including strong downloads and daily use. The next meaningful evidence is retention, completed tasks and whether users keep the permissions that make an agent useful.
Amazon says it has blocked Meta’s Muse from shopping on Amazon.com, turning AI shopping into a question of retailer permission, identity and terms of use rather than model capability alone.
Google says CC is becoming a family AI agent with shared memory and up to six users. The bigger shift is permissions: who can share, who can act, and who can revoke access.
Meta launched Muse on Sept. 8 and brought it to iPhone and Mac web access by Sept. 17. The key story is its permission model: a separate system decides what the agent may do.
GPT-Live-1 is most useful when a user keeps talking while a backend checks or changes something. The real decision is how much control a team needs over context, approvals and what the caller hears.
Anthropic’s redesigned Projects beta coordinates parallel Claude Code sessions. Access starts with selected Pro and Max users, while cloud execution, merge conflicts and faster usage consumption define the initial limits.