Affiliate links on Android Authority may earn us a commission. Learn more.
June 12, 2025
If you’ve ever shared something personal with Meta’s AI assistant, you might want to double-check who saw it. A serious misunderstanding with the functionality of the Meta AI app is causing people to publish deeply private conversations inadvertently on a public feed.
A thread from X user Justine Moore highlighted the issue and shared some examples. The user had seen confessions of affairs, medical questions, legal dilemmas, and tax records — all visible to anyone browsing the app’s Discover tab. Within a few minutes of signing up and logging on, I saw the same sort of thing.
Meta launched its standalone AI app last year. The assistant, which now runs on Llama 4, is integrated into Facebook, Instagram, WhatsApp, and Messenger, and is designed to answer questions, generate images, and help people with everyday tasks. The standalone app also features a Discover section, which is a TikTok-like feed of publicly shared conversations that people can scroll through and interact with.
Many people seem to think the Share button saves the conversation privately, not that it blasts it to a public feed. There’s a small disclaimer and a two-step process, but the interface is evidently misleading enough that some people are publishing their chats without realizing the consequences.
One post reportedly included a user trying to write a romantic poem for his girlfriend, while another had someone asking Meta AI for help finding a woman with a “nice rack,” followed by an attempt to “delete my number” when things didn’t go to plan. Ironically, not only did this not work, but it was also shared publicly.
There are ways to make your posts private again and prevent Meta from suggesting your prompts elsewhere, but many argue that people shouldn’t need to go hunting for privacy settings in the first place. While other chatbots like ChatGPT and Gemini only share chats through opt-in links, Meta’s feed makes oversharing feel like a feature. The Discover tab was likely meant to highlight creative use cases like AI-generated images or quirky writing prompts, but it’s quickly become a privacy minefield.