Meta AI searches go public, sparking user privacy concerns

Users interacting with Meta AI, the generative artificial intelligence tool integrated across Facebook, Instagram, and WhatsApp, may be unwittingly exposing their search queries and the AI’s responses to a public feed.

Despite Meta’s assertion that “nothing is shared… unless you choose to post it,” internet safety experts are warning of a “huge user experience and security problem,” as these public posts can be easily traced back to individuals via usernames and profile pictures.

The “Discover” feed on the Meta AI app and website displays a timeline and collage of user-chatbot interactions. While some queries are innocuous, many reveal surprisingly personal and sensitive information.

Examples uncovered include users uploading photos of school test questions, asking for answers, or requesting images of scantily-clad characters. More alarmingly, users have posted intimate medical questions, like how to deal with a rash, inquiries about personal financial situations, legal matters such as tenancy termination formats or tax fraud liability, and even details about ongoing court cases, sometimes including personally identifiable information such as addresses and phone numbers.

Cybersecurity expert Rachel Tobac, CEO of Social Proof Security, posted on X saying that users do not expect their private chatbot conversations to appear on a public social feed, leading to inadvertent disclosures of sensitive data linked to their identity.

Calli Schroeder from the Electronic Privacy Information Center said in an interview with Wired magazine such sharing points to a misunderstanding of how chatbots function and how privacy works with these platforms, emphasizing that “nothing you put into an AI is confidential.”

Meta maintains that users must go through a multi-step process to share content publicly and are warned to “avoid sharing personal or sensitive information.” However, the sheer volume of personal data appearing publicly suggests a disconnect in user understanding.

While Meta AI continues its rapid deployment, with over a billion users across Meta’s platforms, the debate intensifies over data privacy in this new era of conversational AI.