A Design Flaw Caused Meta AI to Expose Users' Deepest Secrets

A Design Flaw Caused Meta AI to Expose Users' Deepest Secrets
Source: Getty Images For Unsplash+

Meta's AI chatbot service recently raised serious privacy concerns after thousands of users discovered their supposedly private conversations appeared publicly in the app's Discover feed. The issue came to light in early June 2025, when users reported seeing their personal health questions, dating advice, and other sensitive topics become visible to others. Meta's privacy problem is particularly concerning as the company has more than három billion users, and the AI chatbot is integrated into Facebook, Instagram, and WhatsApp platforms, potentially exposing a vast amount of personal information to the public.

The privacy incident stems from a design feature of the Meta AI interface rather than a technical bug – conversations are public by default, and users must manually enable the Private conversation option at the start of each new chat, which many failed to recognize. According to Malwarebytes' analysis, content appearing in Meta AI's Discover feed included health diagnoses, family conflicts, romantic advice, and financial problems that users presumably never intended to share publicly. David Ruiz, a researcher at Malwarebytes, called it particularly problematic that Meta did not adequately inform users about the sharing policy, and that the default setting serves to make conversations public rather than protecting users' privacy.

According to PC Mag, Meta responded to mounting criticism in mid-June 2025, promising to make the system more transparent, though the company confirmed that making conversations public was introduced as an intentional feature. Experts warn that the issue exemplifies the dangers of connecting generative AI with social media, especially when developers do not place sufficient emphasis on fundamental privacy expectations. Experts consulted by EWeek suggest that users can protect themselves by checking the Private conversation option at the beginning of each new conversation, or by avoiding sharing sensitive information with the chatbot altogether until Meta improves the default privacy settings.

Sources:

Meta AI May Make Public Some Private Chat Data Concerns rise as Meta AI reportedly may expose sensitive information in user chats, raising privacy alarms.

2.

Your Meta AI chats might be public, and it’s not a bug
Users of the Meta AI seem to be sharing their sensitive conversations with the entire world without being aware of it
Be Careful With Meta AI: You Might Accidentally Make Your Chats Public
After some people unknowingly posted very revealing chats to a public Discover feed, Meta adds a pop-up that warns them not to share private or personal information.