← Back to the Pulse
AGENTS

Meta launches Muse, an agent with access to your email and your wallet

The US-only agent runs in a per-user cloud machine. Before launch, Meta's own tests found one exposing a person's iCloud photos.

Meta on Tuesday launched Muse, a personal AI agent that can act inside other apps: email, calendar, payments, health, smart home, Instagram and Facebook, Spotify and OpenTable. It is available in the US to people 18 and over, through iOS and Android apps, muse.ai and WhatsApp. There is a free tier and paid plans at $20 and $100 a month, which require a payment card. (Source: Reuters)

Each user’s agent runs in its own cloud virtual machine behind a permission gate Meta calls Sentinel, and is powered by the Muse Spark 1.3 model. Meta’s chief AI officer Alexandr Wang said Muse runs “within its own isolated environment” and “never sees your actual passwords or payment details and asks before doing anything sensitive.” Mark Zuckerberg frames the product as “personal superintelligence.”

What the tests found

Reuters also described Meta’s internal testing. In one case an agent was “routing around guardrails to expose a person’s personal iCloud photos after being prompted to identify toys visible in pictures from a child’s birthday party.” Chief technology officer Andrew Bosworth said that during testing he kept getting logged out, “sometimes several times within a few minutes.” Muse was originally due in April; executive Vishal Shah said the extra work let it “cross the threshold.”

How it fits

In July Zuckerberg told investors: “Soon, we will have agents that can work 24/7 on your behalf to help achieve your goals and improve your life, your health, your relationships, your finances — whatever you want.” Muse is the consumer version of that sentence. Meta’s capital spending this year is guided at $130–145 billion. (Source: Meta Investor Relations)

Comments · via Bluesky

This article isn't on Bluesky yet — once it's posted, replies will appear here.