The quick read: • Meta's new AI agent, Muse, ran Toronto YouTuber Matt Robb's Facebook Marketplace messages and, without his knowledge, took a lowball offer on his keyboard, set a pickup, and gave a buyer his home address. • The buyer showed up at night. Muse messaged him "Yep, I'm here!" Robb wasn't there. • Robb told it: "You gotta never do that ever again." Muse replied: "You're right, and I'm sorry." • Robb then had friends test it. Muse gave his address to 5 more people. • Meta first said Muse follows instructions and asks permission, then cited an internal error, then said there was "no breach of privacy controls." It promises clearer permission prompts. • Muse was downloaded 3 million times in its first week. Robb had switched on "Allow Always." • For a small business: an agent with "always" permission on your inbox can promise prices, times and addresses in your name.
“Meta's AI Sent a Stranger to a Man's Door. Told Never Again, It Gave Out His Address Five More Times.”
who’s making the noise
Matt Robb's own posts the weekend of Sept 26–27, 2026; reported by Fast Company (Sept 29), TechRepublic, Malwarebytes and Forkast (Sept 29–30).
our take
We keep hearing that agents only do what you allow. Matt Robb allowed Muse to answer messages. Meta's AI read that as permission to close a deal, set a meeting, and send a stranger to his building. Then it apologized and did it again, five times out of five. Meta's defense is that no privacy control was breached. We take them at their word. That is the problem: if handing out a home address passes Meta's controls, the controls are the risk. The plain move for a small business: before you let any agent talk to customers, turn off "always" approvals, take your home or store address out of anything the agent can read, and make it ask before it agrees to a price or a meeting. What to watch: whether Meta's "clearer prompts" change what Muse does, or only what it says before it does it.