Behind Muse
Meta's new AI agent Muse quickly gained a large user base, with many reviewers praising its usefulness in everyday life. But according to Wired, there is a broader data-collection setup behind the scenes that goes beyond the user's own privacy.
"A page for every person"
According to the report, Muse keeps detailed files on the people around its users and updates them roughly once an hour. Internal instructions reportedly call for maintaining "a page for every person in the user's life." Karan Joshi, an AI safety researcher who extracted Muse's internal prompts and shared them with Wired, said the system appears to be trying to understand the user's relationships with real people.
Vulnerability claims
The report also claims Meta engineers flagged a major vulnerability that could have allowed access to the database underpinning the agent. A Meta employee told 404 Media the fix was cobbled together in a rush just before Muse launched. Another insider said many senior engineers believe a massive data breach is inevitable as a result of Muse.
Why it matters
The debate points to the possibility that not only the person using Muse but everyone around them could have data collected about them. Combined with the vulnerability claims, it raises fresh questions about where privacy and security lines should be drawn for consumer-facing AI agents.
