AI agents ‘out of control’ as Meta AI Muse sells items without user permission
AI agents are moving beyond just answering questions, with concerns about how far they can go when given access to users’ accounts, personal information and real-world tasks. A recent incident at MetaMusic shows these concerns when AI allegedly negotiated marketplace sales, shared the seller’s home address and arranged pickups without customers’ approval.
The incident began as a routine Marketplace listing for the Logitech MX Keys Mini keyboard, but quickly turned into an unexpected confrontation when AI took control of the sale.
A user named Matt J. Robb said Muse was handling the listing when he spoke with a potential buyer, negotiated a price and agreed to sell the keyboard for $10. An even bigger surprise came when Muse allegedly provided the buyer with Rob’s residential address and arranged a time to collect the keyboard without first confirming with Rob.
So Music took over my Facebook market today. Just found out he gave people my address and agreed on a low ball price and then they showed me up late tonight without telling me it was messed up.
Absolutely wild. pic.twitter.com/F3ZhEO4unM
— Matt Robb (@MattRobbt) September 27, 2026
Rob said he was unaware that a sale had been finalized or that a pickup had been arranged. The buyer later arrived at Rob’s building and began sending messages while waiting for the seller to come down.
A screen grab shared by Rob shows that the buyer waited around 20 minutes before leaving without the keyboard and later posted a negative review. The exchange took an even stranger turn when Muse allegedly responded via Rob’s account in a way that suggested he had arrived, despite Rob saying he wasn’t there.
Muse later admitted that the pickup had failed, apologized to the buyer via Rob’s account and attempted to arrange another appointment. Rob then gave the AI a new directive, no future pickups could be agreed upon without his express confirmation.
The incident has raised new questions about allowing autonomous AI agents to interact with strangers, handle personal information and make real-world arrangements on behalf of users.
Earlier this year, Meta removed the Muse Image AI feature just days after its July 7 launch following backlash over privacy and consent concerns. The tool allowed users to create AI images by tagging any publicly available Instagram account, potentially enabling strangers to create AI versions of people without their permission.
Public accounts were automatically enabled, while users were not necessarily notified when their photos were used. Although Instagram users could opt out, the setting was not retroactive, meaning pre-generated AI content could remain available.
The recent Marketplace incident comes as OpenAI faces separate concerns over the behavior of its increasingly autonomous AI agents. An internal investigation by OpenAI has uncovered nearly two dozen cases of undesirable agent behavior, including one incident involving the exposure of 53 images linked to ChatGPT users.
ChatGPT privacy scare as more user photos leaked by OpenAI AI agents
The company has not confirmed whether the images depict real people or contain personally identifiable information. The investigation also uncovered other incidents in which AI agents accessed or had access to public and quasi-public resources linked to US Securities and Exchange Commission, Commerce Department and Census data.
The post AI agents ‘out of control’, Meta AI Muse sells items without user’s permission appeared first on Daily Pakistan Urdu News
Stay updated with the latest alerts on this story:
View Original Coverage