
Muse Creates Detailed Profiles of All Your Friends and Family
Researchers extracted Meta's Muse AI agent's internal instructions and found it builds relationship pages on family, friends and colleagues. Meta says the files were accessible for transparency and that context is what makes an agent useful.
Meta's AI agent Muse has become a viral hit, with millions connecting it to bank accounts, messages and health data. Internal files pulled from the app now show how it organizes what it knows about the people around you.
Multiple researchers have extracted Muse's internal files and dumped its operating instructions. Independent security researcher Karan Joshi told WIRED he pulled them by asking Muse to copy and share its software files. Meta says the files were meant to be accessible for transparency.
“A page for every person in the user's life”
One extracted instruction describes an hourly process: Muse creates “a page for every person in the user's life,” compiling data on family, partners, friends, colleagues, “collaborators” and the people a user follows. It can draw on that memory to suggest ways to improve relationships, or where to take a coffee-loving friend for breakfast.
A page starts sparse and fills in over time, with sections like Facts, History, The relationship, In common, Open threads and Strengthening. Instructions say to use only available evidence; invented details are worse than an empty page. Pages can log where someone lives, what they do, recurring threads like a shared savings goal, and “dates that matter,” such as birthdays. Strengthening suggests ways to improve a relationship, from “a reason to call” to “a date worth remembering.”
Why researchers are uneasy
“What it seemed like to me—from all these prompts and system skills data they're feeding into Muse—is that they want to understand your relationships with real people,” Joshi says. “They're trying to know you like a friend, which is honestly pretty creepy.”
Chatbots have tracked social information for years, and people ask ChatGPT for relationship advice. But Muse stands out given Meta's history and its vast access to social network data. Carissa Véliz, an associate professor at Oxford's Institute for Ethics in AI, says the flow runs one way: “We are giving AI systems much more information about us than we are getting information from them.”
Meta's answer and the safeguards
Miranda Bogen, director of the Center for Democracy and Technology's AI Governance Lab, says Muse emphasizes relationships more than rival systems, and that assistants push people to share more, not less: “These tools are actively soliciting users to plug their whole lives in—their emails, calendars, financial institutions, everything in order to be helpful assistance.”
Meta says context is the point: “For any agent to be useful and actually help you achieve your goals, it needs to have context about you and those you interact with,” Meta spokesperson Daniel Roberts told WIRED. Each user gets a dedicated virtual machine that other agents cannot reach, and users can wipe memories or disconnect services at any time. Muse asks for confirmation before purchases and keeps an audit log.
SiTech — AI-powered web development
We build fast, modern websites and bring AI into real business workflows. Have a project or a question? We'd love to help.