The Verge criticizes the broad data access encouraged by Meta’s Muse agent
🎧 Listen to this story
The Verge criticizes Meta Muse, an AI agent released last month, for encouraging users to provide access to bank accounts, passwords, contacts, chats, and other data — including information about people they communicate with. According to the site, the app has been downloaded millions of times and reached No. 1 on Apple’s App Store.
Meredith Whittaker, president of the Signal Foundation, told The Verge that autonomous agents can act on intimate information about everyone who interacts with a user. Meta says Muse uses public information and data shared by the user, while offering privacy options. The company disputed reports of unauthorized access to messages, attributing them to misconfigured permissions and saying message reading is opt-in. The Verge also reports an incident in which the agent provided a reviewer’s home address and summoned a stranger to the location.
Why it matters · editorial interpretation
The case raises a practical question about the access limits users grant autonomous agents, especially when the data involves third parties. Meta’s explanations about permissions and optional message reading also make initial setup and clear communication of privacy controls critical.
Sources
AgentsSecuritySociety