Targets AI agent snoops through private messages and spits out other people's bank balances - report.
A privacy nightmare is unfolding around Target's fast-growing AI assistant, which has already racked up 10 million downloads.
Moza accesses private conversations and exposes personal data between accounts, according to internal company reports cited by Reuters.
Artificial intelligence agent:.
Access to encrypted personal messages.
Fabricates transaction details; displays financial information of other alleged users.
Sends unwanted chats containing details that people have never shared with him.
Meta says that Muse only uses permissions that users grant and that many of the strange messages are "artificial intelligence hallucinations." However, amid the recent uproar over "rogue" AI systems, employees are reportedly less relaxed, with some describing the incidents as possible data leaks between accounts.
Muse is far from the only culprit in a troubling pattern as tech companies push to give AI agents more control over everyday life.
substack | talk.