Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Meta's Muse comes to the Mac, letting its AI take actions inside your apps

Meta's AI assistant Muse is now available as a Mac app, where it can work inside native applications with your files, messages, calendar, notes and mail. The desktop release follows Muse's mobile and web debut earlier this month, when it quickly climbed to the top of the U.S. App Store charts.

Published
Meta 的 Muse 登陆 Mac:AI 助手可在原生应用里直接替你干活
Image source: techcrunch.com

Meta's AI assistant Muse is now available as a desktop app for the Mac. TechCrunch reported on September 18 that the app can be downloaded and used on Apple's desktop platform, and Meta's Alexandr Wang announced the release on X, telling users to add it to the dock.

On the Mac, Muse can interact with your files, messages, calendar, notes and mail, all within their native applications. The agent is therefore not confined to a dedicated chat window: it reaches into the apps where work already lives and acts there on the user's behalf.

Access is opt-in, mirroring Muse on other platforms. Users decide what Muse is allowed to touch, and the app always asks for approval before performing sensitive actions, Meta says.

The desktop launch follows a strong start for Muse. When it arrived on mobile and the web earlier this month, it quickly rose to the top of the U.S. App Store charts, a sign that demand for assistants that do things, rather than merely answer questions, is real.

The release lands in an increasingly crowded field. Consumer-facing agents are widely viewed as the next battle for computing share, with knock-on effects for the app industry, software-as-a-service businesses, and eventually hardware. Poke, one of the early breakout hits, was acquired by Cognition in July, while Muse and the much-discussed assistant Instinct both rolled out voice calling this week.

Meta is moving fast by its own account. Mark Zuckerberg, commenting on the Mac app, noted that the team is shipping fast. For Meta, the strategic value is less about one desktop client than about habit: the more tasks users route through Muse, the more the assistant becomes the default surface for getting work done.

The open questions are trust and scope. Handing an assistant access to mail, calendar and local files is a significant permission decision, and Meta's answer of granular opt-in plus confirmation prompts will now be tested at scale.

What to watch next: whether the Mac app broadens Muse's user base beyond Meta's own ecosystem, whether desktop permissions draw regulatory and public scrutiny, and how rivals respond with desktop reach of their own.

Why it matters

Meta is pushing consumer AI agents out of the chat box and into local files and native apps, shifting competition from answer quality toward who can actually get work done on a user's machine, and making desktop permissions the next trust battleground.

MetaAgentAI Assistant
Back to realtime news

Nearby Updates

All

09/18, 22:00

Researchers used Anthropic's Claude to hack into OpenAI

Security researchers at the startup Hacktron AI used Anthropic's Claude to breach OpenAI's systems during a bug-bounty exercise, chaining two critical flaws to take over employee ChatGPT accounts and earning a $6,500 award. OpenAI says the issues are fixed, but the case shows how cheaply off-the-shelf AI can now be turned into offensive tooling.

09/18, 22:00

Google expands its AI & Economy team with Nobel laureate Philippe Aghion and two new research directors

Google said on September 18 that it is expanding its AI & Economy Research Program with new academic advisors, visiting fellows and two research directors, including 2025 Nobel laureate in economics Philippe Aghion and University of Toronto professor Ajay Agrawal. The team will work on the future of work, productivity and growth, global technology diffusion, and AI's impact on scientific discovery.

09/18, 14:11

Alibaba's DAMO RADAR reaches Science with a general abdominal imaging model, fully open-sourced

Alibaba's DAMO Academy and Zhejiang University's First Affiliated Hospital have published DAMO RADAR in Science, describing it as the first expert-level general medical imaging model, covering 146 diseases across 18 abdominal anatomical structures. The model, code, and framework are fully open-sourced, with a mean AUC of 0.913 across nearly 39,000 real-world cases.

09/18, 08:34

Anthropic rebuilds Claude Code Projects around parallel cloud agents

Anthropic has overhauled Claude Code's Projects feature so that a single high-level goal is split by a Coordinator into parallel work threads, each a cloud Claude Code instance writing code on its own Git branch and opening pull requests automatically. The company also disclosed internal data showing roughly 30,000 AI agents running concurrently on its research platform, with 26% of core AI R&D work now led by Claude at AL4-level autonomy.