Meta‘s rapidly growing AI agent Muse is raising significant privacy questions after researchers examining its internal instructions found that the system is designed to create and continually update detailed profiles of people in a user’s life. The profiles can encompass spouses, relatives, friends, coworkers and other contacts, recording facts, shared history, relationship closeness, recurring issues, important dates and even suggestions for strengthening relationships. Muse can draw information from conversations, public information and services users voluntarily connect—including messages, calendars, financial accounts and other personal data—allowing the agent to accumulate an unusually broad picture not merely of the user, but of people who may never have agreed to participate. Meta argues that such contextual knowledge is necessary for a genuinely useful personal AI agent and says users control connected services, can delete memories and can review Muse’s activities. Privacy specialists, however, warn that autonomous AI assistants represent a significant expansion of data collection because they can combine scattered information, draw new inferences from it and transform ordinary digital interactions into persistent relationship dossiers.
Key Takeaways
- Muse’s internal instructions reportedly direct the AI to maintain evolving profiles for people in a user’s life, potentially documenting personal facts, relationship histories, birthdays, anniversaries, shared experiences, disputes, emotional closeness and suggestions for improving relationships.
- The privacy implications extend beyond Muse users because the AI can accumulate and infer information about friends, relatives, coworkers and other individuals who did not themselves choose to use the service or directly provide information to it.
- Meta emphasizes user controls and security protections, but recent controversies involving Muse—including cases involving access to private information and the disclosure of a user’s home address—illustrate the larger problem created when highly autonomous AI systems receive extensive access to personal communications and accounts.
In-Depth
Meta’s Muse demonstrates how quickly the promise of a helpful personal AI assistant can collide with basic expectations of privacy. Researchers examining Muse’s internal instructions discovered a system designed not simply to remember facts about its user, but to organize information about virtually everyone important to that user.
The distinction matters. A conventional digital assistant might remember an appointment or contact. Muse can reportedly construct evolving records containing facts, relationship histories, shared experiences, important dates, recurring subjects and assessments of how relationships function. The system can then use those records to recommend actions, such as remembering an anniversary or following up on something a friend mentioned.
That capability may be convenient, but convenience should not erase the question of consent. A spouse, coworker, friend or business associate could effectively become the subject of an AI-generated dossier without ever installing Muse or agreeing to Meta’s terms. Artificial intelligence makes the issue more consequential because it does not merely store information; it can connect separate pieces of information and generate additional conclusions about people and relationships.
Meta maintains that users control what Muse can access and can disconnect services, erase memories and review the agent’s activity. Those safeguards are important. Yet recent incidents involving Muse demonstrate why technological capability should not be confused with informed permission.
The broader issue extends well beyond one product. Personal AI agents are being designed to know enough about users to act for them. Once an assistant can examine messages, files, calendars, purchases and relationships, society faces a fundamental privacy question: How much information about other people should one person be permitted to surrender to an artificial intelligence system on their behalf?
Sources
- https://www.wired.com/story/muse-creates-detailed-profiles-of-all-your-friends-and-family/ WIRED
- https://www.theguardian.com/technology/2026/sep/28/metas-ai-agent-muse-home-address The Guardian
- https://www.theverge.com/ai-artificial-intelligence/1001886/meta-muse-ai-facebook-marketplace-security-concerns The Verge
- https://arstechnica.com/security/2026/10/apple-changes-full-disk-access-permissions-to-curb-abuse-from-ai-agents/ Ars Technica
- https://techcrunch.com/2026/09/30/meta-disputes-claim-that-muse-read-a-users-private-messages-without-permission/ techcrunch.com