Imagine asking an AI a private question or for help with something personal, only to find that conversation appearing online for others to see. This surprising situation might be happening to some users of Meta AI, as prompts and generated responses are sometimes posted onto a public feed without users fully grasping the implications. This could mean unintentionally sharing sensitive queries with the world, from help with exams to exploring personal topics or generating specific images. It’s a stark reminder to be mindful of where and how you interact with AI tools, especially within social platforms.
Contents
The Public Side of Meta AI
Meta AI, integrated into platforms like Facebook, Instagram, and WhatsApp, and also available as a standalone web experience or app (in the US), includes a feature called the “Discover” feed. This feed is designed as a place for users to share and explore interesting ways people are using the AI.
The crucial point is that conversations you have directly with Meta AI in a private chat are, by default, private. However, Meta allows users to explicitly post interactions from their chats to this public Discover feed. When a user chooses to post, a pop-up message warns that “Prompts you post are public and visible to everyone… Avoid sharing personal or sensitive information.”
A close-up shot shows a woman's hands holding a green smartphone, interacting with an AI chatbot interface.
Despite this warning, the nature of some publicly shared posts suggests that users might not fully understand that their interactions could become visible to anyone, or that these public posts might be traceable back to their social media profiles via usernames and profile pictures.
Surprising Examples Found in the Public Feed
The public feed has revealed some unexpected posts. For instance, some users have uploaded photos of school or university test questions, asking Meta AI for the answers. One publicly posted chat was titled “Generative AI tackles math problems with ease,” showcasing this use case.
Other examples touch on deeply personal subjects, like one conversation where a user was exploring questions around their gender identity and potential transition.
There were also instances of users asking the AI to generate images of characters, including anthropomorphic animals, wearing minimal clothing. In at least one case, a request to generate an image of an animated character in underwear was traceable back to a user’s Instagram account through their username and profile picture. These examples highlight the range of queries users are running and the potential privacy exposure if these are unintentionally made public.
What Meta Says and Expert Concerns
Meta maintains that users are in control of what they share. In a press release announcing the standalone Meta AI product, the company stated, “You’re in control: nothing is shared to your feed unless you choose to post it.” Users can also adjust their account settings to manage privacy.
However, cybersecurity experts point out a potential disconnect between how users expect a private chat tool to function and the social media-like public feed feature. Rachel Tobac, CEO of Social Proof Security, noted that this mismatch between user expectation and reality creates a “huge user experience and security problem.” She explained that people typically don’t anticipate their interactions with an AI chatbot appearing on a public feed linked to their identity, which can lead to the inadvertent sharing of sensitive details.
It’s important for users to be aware of these nuances. While private chats remain private, posting interactions to the Discover feed makes them public. If you’re using Meta AI, especially the standalone version or considering sharing, double-check the privacy settings and be cautious about the information or queries you choose to post publicly.
For more on AI and privacy concerns, read about calls for Meta to crack down on ‘nudify’ apps and how WhatsApp defended its ‘optional’ AI tool that couldn’t be turned off.
