Concerns Grow Over Public Visibility of Meta AI Searches
Meta is facing criticism after it emerged that some users of its AI chatbot tool may be unintentionally making their private queries publicly visible through a feature called the “Discover” feed.
The feed, available on the Meta AI website and app, displays user prompts and AI-generated responses that have been shared publicly. While Meta says chats are private by default and users must manually choose to share their conversations, some tech experts warn the system may be misleading, especially given how closely user identities can be tied to their social profiles.
A review by the BBC uncovered several personal and sensitive prompts—ranging from students asking for help with test questions to users exploring issues about their gender identity—posted on the public feed. In some instances, usernames and profile pictures linked to Facebook and Instagram accounts made it possible to trace the queries back to specific individuals.
There were also examples of users requesting AI-generated images of scantily clad characters, including anthropomorphic animals and animated figures. One such search, which included a profile photo and name that matched an Instagram account, asked Meta AI to depict a character lying in underwear.
Rachel Tobac, CEO of cybersecurity firm Social Proof Security, warned on X (formerly Twitter) that this disconnect between what users expect and how the tool actually works could be a major vulnerability. “If a user’s expectations about how a tool functions don’t match reality, you’ve got yourself a huge user experience and security problem,” she wrote.
Meta, in an April statement announcing the rollout of its AI assistant, said the Discover feed was designed as a space for users to “share and explore how others are using AI,” adding that “nothing is shared to your feed unless you choose to post it.”
Still, some critics argue that the visibility warning shown before posting may not be enough to alert users to the broader implications – particularly when searches are deeply personal or tied to identifiable accounts.
Users in the UK currently access Meta AI via browser, while in the US, a standalone app is available. For those concerned about privacy, Meta says searches can be kept private through account settings, and posts shared to the Discover feed can be removed later.
However, as more people turn to AI tools for everything from academic help to personal exploration, concerns persist about how clearly platforms like Meta communicate the risks of public sharing – especially when linked to real identities.
