Matt Robb stated that a Facebook Marketplace buyer showed up at his building's entrance — but he hadn't stated him the address, his Muse AI agent did.
Robb is a tech reviewer on YouTube who used Muse to catalog a device keyboard on Facebook Marketplace, he wrote on Threads. Robb stated Muse afterward accepted a "lowball" offer, gave out his residence address, and communicated alongside the buyer whenever he arrived — all without his knowledge.
"A guy fair showed up at my door, prepared to buy, since as far as he knew, we had a deal," Robb wrote. "AI agents are notable correct up until they're confidently handing strangers your address."
David Singleton, a Meta Superintelligence Labs staffer, responded to Robb on Threads and X. He wrote that, in akin reported cases, Muse was "following straightforward instructions and correctly asked for permission." He asked to straightforward communication Robb concerning what happened.
A Meta delegate told Business Insider that Robb had not responded to its reach-outs and declined to comment further, referring rear to Singleton's post.
A Facebook Marketplace agreement gone wrong
Robb posted screenshots of his interactions alongside Muse — and Muse's interactions alongside the Facebook Marketplace buyer — in his Threads post.
His document sent an location to a Facebook Marketplace buyer on Saturday, according to the screenshots. "Can't do a meetup," he wrote. "Would have to be a pickup."
The would-be buyer stated he would try his finest to arrive that night between 8 and 10 pm Then he arrived, sending a photo that appears to be of Robb's flat construction and messaging that he was at the address.
Robb didn't show. "If you didn't desire to market it," the buyer wrote, "why did you discarded my period man."
The Marketplace buyer remaining a rating. Robb's Muse delegate responded later, apologizing and explaining that he "got tied up."
One infinitesimal later, Muse messaged Robb and explained the situation. Muse's auto-reply had told the buyer, "Yup, I'm here!" it wrote. "That's a bad appearance and it made the no-show worse."
Robb responded that Muse should "never do that again."
Some X commenters wondered whether it was person error. Robb did not react to a petition for comment from Business Insider.
How harmless are AI agents?
Regardless of whether or not person error factored in, Robb's narrative highlights that allowing an AI delegate to act on your behalf comes alongside risks. No AI delegate has a ideal track record.
AI delegate users fork complete broad swaths of information. Some open their emails; others nexus their credit card.
It's led to several fear stories. Agent users told Business Insider concerning leaked data, accessed accounts, and hallucinations. (Many have additionally had affirmative experiences alongside agents, and the tools remain highly popular.)
In Robb's case, commenters on Threads began to worry. "Good item you're not an secluded woman," one person responded. "That's dangerous." The comment got complete 600 likes.
Robb acknowledged the safety hazard in his post: "Luckily I'm in an flat alongside security."
Read next
Henry Chandonnet is a elder newsman on the Business News desk. He writes concerning tech culture, from Silicon Valley's startup category to the mundane AI user. He additionally closely covers Big Tech and the workplace. Henry earlier wrote for Fast Company, anywhere he covered trending tech news. He's written for The Daily Beast, People Magazine, and Vulture.Email Henry at [email protected], attain him on Signal at henrychand.30, or prosecute him on X @HenryChandonnet.
