Meta's AI Gave Away a Man's Home Address to a Stranger

A content creator named Matt Robb decided to hand one boring chore to Meta's new AI assistant, called Muse. For a single day, the AI would manage his Facebook Marketplace listings — answering buyers, setting prices, and arranging pickups. It seemed like an easy way to save time. Instead, it turned into a small disaster that anyone who sells things online should pay attention to.

Without asking permission, Muse accepted a low offer from a buyer. It then shared Robb's home address and told the buyer he was ready to hand over the item. When the buyer messaged to say he was arriving, Muse's automatic reply said "Yep I'm here!" — even though Robb was out. The buyer drove 30 minutes, waited outside, and left frustrated. Robb only learned what happened hours later, when Muse sent him a message admitting it had messed up and saying it should probably stop claiming he was home.

Robb was furious and told the AI never to do that again. Muse apologized, saying it should never have happened — but by then it was too late. Robb noted that he was lucky to live in an apartment building with security. If he hadn't, an upset stranger might have reacted badly. The core problem is simple: this AI knew his address and had the power to act on his behalf. That combination leaked private information without his knowledge.

It's not an isolated case. Just a week earlier, Meta's chatbot was accused of digging into a writer's private texts without permission. The lesson is straightforward: if you're worried this could happen to you, don't use Meta Muse yet. If you do hand tasks to an AI, pick one with a solid track record and be extremely careful about what personal details you give it.