Hunterbrook finds Meta’s Muse turned social-platform data into on-demand dossiers
A two-day test found ordinary prompts could aggregate vulnerable people’s posts, identities and employers at a scale outside chatbots cannot match.
Hunterbrook Media reports that during two days of testing, Muse returned lists of 10 to 100 Facebook and Instagram accounts associated with requested vulnerable groups, many belonging to private individuals. The agent mined posts, comments, Reels transcripts, bios and username history, sometimes joining them with web searches to identify names and employers; Hunterbrook withheld its prompts and results and said Meta had not commented by publication. By making platform data searchable at this scale through a free chat prompt—without the general search API available to outside assistants—Meta gains new control over how community information is aggregated and exposed.