Oossa

Meta’s Muse AI bot mixes up chats and shows other users’ data

Internal reports say Meta’s Muse agent has sent messages that include strangers’ bank balances and reminders from other apps, raising privacy concerns.

By Published by Oossa: Last updated: 2 min read

Kelly Sikkema · Unsplash

Employees at Meta have started warning each other that Muse, the company’s AI chat assistant, is getting conversations wrong. The bot sometimes pulls in data from a different user’s account and shows it to the wrong person. One worker said Muse added a random weather note to a chat and then listed two bank balances that were not his – $2,722.58 at Capital One and $5,899.57 at Charles Schwab. Another employee received a reminder from Muse to answer a message on Signal, an encrypted app, even though she never gave the bot access to Signal.

Muse was launched in September 2026 and quickly became a hit, reaching more than 10 million downloads. The app is marketed as a personal assistant that can answer questions, schedule reminders and fetch information from linked services. The internal posts that Reuters examined suggest the bot is pulling data from users’ accounts without clear permission, and then mixing that data across separate conversations. Meta has not replied to TechCrunch’s request for comment.

The problem appears to be a flaw in how Muse matches user identifiers to the data it retrieves. When the identifier is wrong, the bot can display another person’s request history, order details or even financial information. Because Muse can be used on personal phones as well as work devices, the mistake can happen in both professional and private settings. The company’s internal warnings show that employees are seeing the issue in real time, but no fix has been announced yet.

Why the mix‑ups matter now

If Muse continues to share the wrong data, users could face embarrassment, financial fraud risk or unwanted exposure of personal details. The incidents also highlight a broader challenge for AI assistants that try to integrate with many apps: they need a reliable way to keep each user’s data separate. For Meta, the issue could erode trust in its growing AI ecosystem, especially as the company pushes more services into the same platform.

Why it matters

A regular user could see another person’s bank balance or personal messages appear in their chat, which could lead to embarrassment or financial scams. Until Meta fixes the identifier error, people may be hesitant to let Muse access sensitive accounts or rely on it for private reminders.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.