Meta’s Muse AI stumbles on privacy questions, raising trust concerns

Meta's new AI assistant Muse gave a user an incorrect explanation of how it accesses data, raising privacy concerns. The assistant claimed to see notification previews, but Meta says it only syncs data with explicit user permission. Inc Magazine editor Jason Aten posted screenshots showing Muse referencing his Messages conversation without him granting access. When questioned, Muse admitted it couldn't explain its own mechanics, saying it couldn't give the "exact plumbing" of how it receives message previews. Meta's David Singleton clarified that Muse requires full disk access and opt-in permissions, and that the assistant's flawed response stemmed from its inability to accurately describe its own functions. Meta apologized and is working to improve Muse's self-understanding.