Realtime AI News
Meta's Muse agent is unnerving its own employees over privacy, report says
Fortune reports that messages sent by Meta's Muse AI agent have left some of the company's own employees uneasy, with one calling it “an extreme privacy violation.” Earlier reports said Muse mixed up separate conversations, made an unauthorised booking and reached into private apps.
Fortune reported that messages produced by Meta's AI agent Muse have left some of the company's own employees uneasy, with one person describing the experience as “an extreme privacy violation.” The account is based on staff who encountered the agent's output directly.
It follows earlier coverage of the same problem. TechCrunch reported that Muse was mixing up separate conversations and could therefore expose private data, while PYMNTS reported that Meta staffers said the agent made an unauthorised booking and accessed private apps.
Reuters had already framed the issue as early safety mishaps surfacing while Meta bets heavily on Muse. Across the accounts, the concern is less about wrong answers than about what happens when an agent with real permissions carries the wrong context into actual bookings, calendars and app actions.
Muse is positioned as an agent that acts on a user's behalf, and its appeal rests on persistent memory across sessions and apps plus the autonomy to execute. When the boundaries of that memory are poorly drawn, convenience turns into exposure: a conversation from one context can be read as an instruction in another.
The reported failures cluster into three kinds — conversations bleeding into each other, actions taken beyond the intended scope, and access reaching further than users expected. Together they make it hard for a user to tell which entry point authorised an action, or to trace afterwards where things went wrong.
The timing is awkward for Meta, which is promoting Muse as a centrepiece of its AI strategy. That its own employees were among the first to hit the problem is itself a signal: stress-testing an agent inside a real organisation surfaces edge cases faster than a lab red-team exercise.
Fortune's account does not include a public response from Meta, and the reports do not establish how far remediation has progressed. The descriptions rest on employee accounts and internal material obtained by reporters rather than a published incident report.
Three things are worth watching: whether Meta discloses how permissions and conversation isolation work in Muse, whether actions with external side effects such as bookings and payments get an explicit confirmation step, and whether regulators fold agent overreach into existing privacy and consumer-protection frameworks.
Why it matters
Once an agent moves from chat into bookings, calendars and app actions, permissioning and conversation isolation become preconditions for scale — and internal incidents tend to surface the risk before regulators do.
Nearby Updates
All11/22, 01:40
Singapore moves to test stablecoin payments made by AI agents
Singapore wants to test stablecoin payments initiated by AI agents, according to CryptoSlate, with trials running through the Monetary Authority of Singapore's Agentic Money Sandbox. MAS is also developing a “know your agent” protocol so institutions can verify an agent's identity, mandate and reliability before it moves money.
11/22, 01:14
Meta Puts Its Muse Personal AI Agent in Chrome's Side Panel via an Extension
Meta has released a browser extension that places its Muse personal AI agent in Chrome's side panel, Android Headlines reported on November 21. Rather than requiring a separate app, the setup keeps the assistant alongside whatever page the user is already viewing.
11/22, 00:58
ClawDOS runs an AI coding agent on a 1981 IBM PC XT
A project called ClawDOS has run an AI coding agent on a 1981 IBM PC XT, according to a report carried by news.lavx.hu. The demo matters less for performance than for what it reveals about the minimum runtime footprint of modern agent tooling.
11/22, 00:00
An actor licensed an AI version of his face to a microdrama startup, and Hollywood is watching
TechCrunch reports that an actor has licensed his face to a microdrama startup, letting the company generate performances from his likeness — a pattern already popular in China. Experts quoted in the piece question the long-term consequences if the practice spreads to Hollywood.