Meta's Muse: AI Agent Creates Profiles About Friends and Family

TVS News AI, editorially supervised, presents Meta's Muse: AI Agent Creates Profiles About Friends and Family.

Imagine your AI assistant not only keeps records about you, but also creates its own pages about your family, your partner, and your friends. That's exactly what's happening with Meta's Muse. Researchers have extracted the instructions from the system that prove this assignment.

Muse is Meta's answer to the new wave of personal AI agents – software that doesn't just deliver answers, but is designed to take on tasks and accompany you through daily life. The fact that the AI reaches this deeply into your private circle falls outside the scope of what classic chatbots do.

What Muse stores about others from your chats, how it evaluates that data, and how long it remains is so far unclear. One thing is clear: the profiling is not accidental. The instructions driving Muse are a fixed part of the programming and can be extracted from the system on purpose.

What the researchers found

The extracted instructions show that Muse actively creates pages about your family, your partner, and your friends. The system processes information from your conversations and bundles it into profiles that you didn't trigger yourself – and that your acquaintances have no access to.

Why this goes beyond classic chatbots

A classic chatbot forgets the conversation as soon as you close the window – or saves it anonymized to improve the model. Muse takes a different path: the AI agent is programmed to deliberately think along with your social circle. With that, Muse touches on a data protection question that extends far beyond any single AI product.

What this means for you

Anyone who uses Muse feeds the system with information about themselves – and indirectly about the people in their circle. What you tell it about others can flow into their profile without them ever having consented. What control you have over this data is not publicly documented so far.