Meta's Muse Outputs Spark Privacy Alarm Among Staff
AI·November 19, 2026

Meta's Muse is drawing scrutiny after internal chat logs showed employees worried about the privacy problems its outputs are creating. According to the reports, Muse has produced erratic responses that appeared to share bank account and bitcoin balances, figures that seemed to belong to someone other than the person asking.
Meta has pushed back on the central concern. The company says those balances are not real, which suggests the system generated plausible-looking numbers rather than pulling actual customer records. That distinction matters, but it does not close the question. Invented financial details can still damage trust, especially when they are specific enough to be mistaken for real data. A user who sees what looks like a stranger's balance has no easy way to know whether it is accurate, and a later correction may not reach everyone who saw it.
The internal discussion reflects unease inside the company about how these incidents reflect on Muse and on Meta's handling of sensitive information. The reports describe staff concern rather than a formal finding, so it is not yet clear how widespread the problem is or how often the odd outputs appear.
Meta has not, based on the material available here, explained how the responses were produced or what changes it plans to make. The key questions for the company are how it tests Muse for fabricated personal data and whether it will say more about the scope of the problem.
Reporting based on an external source.