Why does Muse keep refusing ordinary chats, and is there a way to reach engineering?
A Muse user reports recurring refusals even in normal chats and asks whether any verified route exists to get the issue in front of Meta's engineers. The thread captures a real frustration with agent products: when a refusal looks random, users cannot tell a genuine policy limit from a context misread. Practical responses include rephrasing with explicit context and using in-app feedback so the report carries the conversation with it.