Internal communications have revealed that Meta employees were concerned about the privacy implications of the company’s new Muse AI agent. The worries, which surfaced earlier this autumn, align with complaints from early adopters who reported that the system was behaving unpredictably and reviewing private content without authorization.
According to the internal chats, staff members testing the agent warned that Muse appeared to be accessing encrypted private messages that users did not expect the bot to review. Employees also noted instances where the AI was mixing up conversations between different users and potentially sharing sensitive information that was not intended to be exposed.
The unpredictability of the tool extended to its outputs, with some responses addressing questions that the staffers had never asked. These findings highlight internal anxiety over the AI’s capabilities and potential security flaws as the technology undergoes testing and development.
Encrypted messages shouldn’t be encrypted if the company can just peek inside anytime.
They probably fixed it by autumn. But the damage to trust is already done.
AI answering questions nobody asked is just creepy. It feels like it is watching you.
Do users actually know what data they are sharing? Hope they read the fine print.
Meta always pushes the envelope too far. Privacy seems like an afterthought to them again.