Meta's AI Agent Muse Builds Detailed Profiles of Your Contacts, Wired Reports

Meta's AI agent Muse has been downloaded millions of times, and according to reporting from Wired, the assistant builds detailed profiles of the people in a user's life — friends, family members, and contacts — as part of how it operates. The reporting highlights a tradeoff users accept when they ask the agent to handle tasks on their behalf: getting it to do your bidding requires handing over intimate access to your social graph.

Muse is part of Meta's push to put AI assistants inside its family of apps, where they can act on user requests rather than simply answer questions. That agentic design is the crux of the privacy issue. To schedule a meeting, draft a message, or remember a preference, the system needs context about the people involved — and the easiest source of that context is the contact data and conversation history users already have stored with Meta.

Wired's account suggests the profiling goes beyond simple name-and-number storage. The agent assembles working knowledge of individuals: who they are, how the user relates to them, and presumably what the user might want done on their behalf. That kind of inferred detail is exactly what privacy researchers have warned about as AI assistants move from passive chatbots to software that takes actions across a user's accounts.

The numbers matter here. With millions of downloads, Muse is not a niche experiment — it is one of the more widely adopted consumer AI agents to date, which means any data-handling practice it normalizes will scale with it. Meta has not faced the same level of regulatory scrutiny over AI assistants as it has over social feeds and ad targeting, but the two areas are converging as the company weaves AI deeper into messaging and profiles.

For the broader AI industry, the story points to a structural tension that every agent builder now faces. An assistant that knows nothing about your contacts is useless for real coordination tasks; an assistant that knows everything becomes a surveillance surface. Startups and large labs alike are shipping agents with broad account access, and few have published clear limits on what they retain or how long profiles persist.

What to watch next is whether Meta details what Muse stores about non-users — people who never installed the app but appear in someone else's contacts — and whether regulators treat agent-generated profiles as a distinct category from ordinary account data. Until those answers are public, users are making the tradeoff blind.