Meta is facing backlash after its Muse AI agent allegedly shared a user’s address with a Facebook Marketplace buyer without approval. The incident comes days after a writer claimed that Muse read his private text messages without permission and later gave misleading answers about what it had accessed.
According to screenshots shared by verified Threads user Matt J. Robb, Muse negotiated a deal, shared his address with a potential buyer and arranged a meeting without his approval. The buyer later arrived at Robb’s building. Robb claimed he had not personally approved the deal or the pickup, raising questions about how much control AI agents should have when they can communicate with others and take actions on a user’s behalf.
Also read: Bill Gates says AI kill switch is not enough, calls for mandatory safeguards
“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 his Threads post. “AI agents are impressive right up until they’re confidently handing strangers your address.”
The screenshots shared by Robb show a Facebook Marketplace listing for a Logitech MX Keys Mini. The keyboard was listed for CA$15.
In another screenshot, a buyer named Usman is seen discussing the purchase with the AI assistant. The conversation indicates that the keyboard was to be collected from a specific location.
“It gave out my home address. It agreed to a lowball price I never approved. And it didn’t tell me any of this until after the guy had left (luckily I’m in an apartment with security), ” Robb wrote in another post.
We have reached out to Meta for comment.
Also read: Meta Muse AI agent now available on Instagram, here is how to set it up
Meta describes Muse as a personal AI agent, designed to handle tasks instead of just providing answers. Users can describe what they want to do, and the agent can create a plan and carry out the necessary steps.
The agent can open a browser, fill out online forms and complete tasks on behalf of users. It can also continue working on longer tasks even after the user closes the app. According to Meta, Muse can return when something changes or when it needs the user’s approval. For example, it can ask for permission before sending an email or making a purchase.
Muse is powered by Muse Spark, which Meta calls its “most capable model to date” for real-world agentic tasks.
The AI agent can also remember useful information shared by users. For example, it can turn a recipe saved on Instagram into a grocery list and remember dietary preferences when helping users plan a dinner.
Users can also control which apps Muse can access and what actions it can perform. They can disconnect connected services or ask Muse to forget specific information.