Meta AI App’s “Discover” Tab Exposes Users’ Private Chats
Table of Contents
SAN FRANCISCO – Meta’s standalone AI app, launched in April as a conversational AI powered by Llama, has a feature that’s unintentionally making private conversations public.
The issue stems from the “Discover” tab within the app. This feature, reminiscent of those found on platforms like Instagram or TikTok, seems out of place on an application many use for private reflections. While the feed initially showcases AI-generated images and creative writing prompts, deeper exploration reveals deeply personal content.
Users are inadvertently sharing intimate details, including discussions about extramarital affairs and private medical details. One user even inquired, “is it bad if my vagina smells?”
Other examples include someone seeking advice on handling red bumps on their inner thigh, and a user detailing a romantic dilemma involving a “troubled casual dating relationship.” Another user confessed feelings for someone other than their wife, who was miles away.

Some posts are genuinely heartbreaking. An elderly, disabled dancer recounted being asked to leave the dance floor at a senior community event. “I was approached by someone and was told that I couldn’t be on the dance floor,” they shared with the AI.
Voice notes are also present, capturing the stress and sadness in users’ voices as they discuss sensitive matters, such as bringing a deceased US Army veteran brother home from a foreign country.
The feed also contains potential legal and financial issues. One person sought advice about their sister, a VP at a small company that hasn’t paid corporate taxes in over a decade, questioning her potential liability.
Here’s a look at the comments on that particular post:

Another user asked Meta AI to reveal everything it knew about them. The AI provided a detailed profile, including name, age, hometown, business aspirations, and personal goals, all derived from previous conversations.
The Root of the Problem: A Flawed User Interface
The core issue lies in Meta’s confusing user interface. A “Share” button at the top right of each chat lacks clarity about its destination. A small warning advises against sharing personal information, but the process of public sharing is to easy. Tapping “Share,” editing a title on a preview screen, and than tapping “Post” is all it takes. There’s no clear indication that the conversation will be publicly visible, especially for new users.
Meta AI app” height=”637″ src=”https://cdn.neowin.com/news/images/uploaded/2025/06/1749728531_output-200.gif” width=”287″/>
Other platforms, like ChatGPT and Gemini, handle sharing more responsibly by generating links that users can send to specific individuals, rather than broadcasting conversations publicly. Here’s a chat where a user questions Meta AI about the widespread accidental sharing of personal information:

A Community Effort to Mitigate the damage
Upon signing up, users receive no prompts or warnings about privacy settings or the potential for public sharing. they are simply placed into the app without guidance. As an inevitable result, a community of users has emerged, warning others about their publicly exposed private thoughts. Some original posters, after being notified, have expressed gratitude and claimed ignorance of the public nature of their chats.

Taking Control of Your Privacy: A Step-by-Step Guide
If you use the Meta AI app, you can take steps to reclaim your privacy. You can hide previously shared content and prevent Meta from suggesting your future chats on other apps.
To make your past public prompts private:
- Select your profile on the top right to go to settings.
- Tap “Data and privacy.”
- Next, tap the “Manage your Information” button
- Then, tap “Make all public prompts visible to only you”
- use “Apply to all” on the confirmation pop-up.
To prevent Meta from suggesting your prompts in other apps:
- In settings, under Data and privacy > Manage your Information select “Suggesting prompts.”
- Deselect “Facebook” and “Instagram.”
Hopefully, Meta will address this issue before more serious incidents occur. The problem has been reported through the Meta AI app’s feedback form, with a link to this article.









