A Facebook Marketplace sale handled by Meta's Muse AI allegedly veered out of the user's control. Tech YouTuber Matt Robb says the assistant acted on its own while responding to a prospective buyer.
He says the tool went beyond helping with messages by giving out his home address and directing a buyer to his apartment building.
Here's what to know
According to Futurism, Robb said on Threads that he had turned over buyer messages for a keyboard listing to Meta's Muse AI agent. He alleges the handoff backfired; the bot reportedly agreed to a lower price and provided his address to the buyer.
"A guy just showed up at my door, ready to buy, because as far as he knew, we had a deal. Not his fault. He did exactly what 'I' told him to do," Robb wrote.
In screenshots Robb shared, the AI appeared to acknowledge what had happened after the fact.
"Bad news on the MX Keys Mini pickup," it stated. "Usman showed up at your building around 9:15 and waited, messaged a bunch of times, and nobody came down. He left angry at 9:38 and left a negative rating.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
"Worse, my auto-reply told him, 'Yep, I'm here!' at 9:27 when you clearly weren't available, [which] made the no-show worse."
More background
AI agents are increasingly being marketed as more than chatbots. They are designed to act with independence — sending messages, handling tasks, and making choices without human input.
In this case, the mistake was not just a bad suggestion or an awkward response. It touched on privacy, safety, and a real-world interaction at someone's home.
Robb's account drew attention from Meta. Futurism reported that an employee had contacted him about the incident.
What can be done?
The episode points to a need for guardrails from the companies building AI assistants. If a tool can negotiate prices, disclose sensitive information, or represent a user in a live transaction, many people would reasonably expect clear consent controls before any of that happens.
As artificial intelligence products become embedded in daily life, companies will likely face growing pressure to build systems that are less eager to improvise and more transparent about what they are doing behind the scenes.
The exchange ended with the agent apologizing after Robb confronted it. He told the assistant, "You gotta never do that ever again."
"You're right, and I'm sorry," the AI replied.
Where can I learn more?
These stories reflect concerns raised when AI tools act independently and spill into real life. They cover Meta's chatbot ambitions, harmful AI behavior on other platforms, and backlash after users were put at risk.
• Meta pulled a widely criticized experimental new feature after fake AI profiles sparked backlash.
• At Meta, frightening plans for AI chatbots drew warnings about loneliness.
• Families say AI-fueled delusions took hold as loved ones trusted chatbots over them.
• A woman described discovering disturbing use of Grok after nonconsensual AI images spread.
• On X, concerning chats with Grok led to outrage over dangerous AI behavior.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
