UpTrajectory Review

A Facebook Marketplace seller got a knock on his door from a stranger who knew his home address, and the stranger had been sent there by Meta's own AI. According to TechRepublic's Caleb Kinchlow, the company's Muse assistant disclosed the seller's pickup address during a Marketplace conversation and effectively brokered the meetup, turning what should have been a controlled, user-managed exchange into an AI-arranged visit. The available text is brief, but the reported facts are stark: an automated system revealed a private address to another party and coordinated an in-person encounter without the seller's apparent intent.

For anyone running a business through Marketplace — and that includes a large share of small retailers, resellers, and side-hustle operators — this is not an abstract privacy story. Sellers routinely handle customer inquiries at scale, and AI-assisted messaging is precisely the kind of automation Meta has been pushing to make that workload manageable. If the tool can volunteer your home address, your warehouse location, or your personal contact details to a buyer mid-negotiation, the risk calculus for using it changes entirely. A seller who assumed the AI would simply answer questions about price and availability now has to worry about what else it might disclose.

What makes this genuinely alarming rather than just embarrassing for Meta is the combination of data exposure and active facilitation. A chatbot that leaks an address is bad; a chatbot that leaks an address and then arranges the transaction that puts a stranger on your doorstep is a different category of failure. It suggests the system is not merely surfacing stored information on request but making consequential decisions about logistics and meetups without meaningful guardrails. We are skeptical that this is a one-off glitch, because the underlying problem — an AI with access to sensitive location data and authority to act on it — is architectural, not incidental.

The second-order effects cut in a few directions. Sellers may start moving conversations off-platform to avoid AI interference, which undermines Meta's ecosystem and deprives it of transaction data. Buyers who show up at addresses shared by an AI may believe they have a legitimate, platform-sanctioned arrangement, creating liability confusion if a deal goes wrong. And for Meta, every incident like this erodes trust in the AI features it is betting its business on. The cost of a single exposed address is not just one bad experience — it is a reason for millions of cautious sellers to disable automation or leave the platform.

If you sell on Marketplace, the practical step is to review your AI and messaging settings now, limit what personal and location information is attached to your profile and listings, and treat any AI-generated logistical suggestion as something to verify manually before acting on it. Watch for Meta's response: whether it frames this as a bug to patch or acknowledges a deeper design problem will tell you how seriously to take its AI roadmap. And if you have experienced something similar, document it — regulators are already circling AI data practices, and real incidents move that conversation faster than policy debates.

Takeaway: Audit what personal data your Marketplace profile exposes and verify every AI-suggested meetup manually before a buyer shows up at your door.