Imagine if your internet search history was suddenly made public for everyone to see. This may be the case for some users of Meta AI, as their prompts and search results are being posted on a public feed without their knowledge. According to an internet safety expert, this poses a significant user experience and security issue, as some posts can be easily linked back to social media accounts through usernames and profile pictures.

This means that individuals may unknowingly disclose sensitive information, such as requesting the AI to generate inappropriate content or help them cheat on tests. It is unclear whether users are aware that their searches are being shared on the Meta AI app and website, although the process is not automatic. When users choose to share a post, a warning message appears, advising them to avoid sharing personal or sensitive information.

The BBC discovered instances where users uploaded photos of test questions and asked Meta AI for answers. One chat was titled "Generative AI tackles math problems with ease", while other searches included requests for images of scantily-clad women and anthropomorphic animal characters. In one case, a search linked to an Instagram account asked Meta AI to generate an image of an animated character in underwear. Additionally, TechCrunch reported cases of users seeking advice on intimate medical issues, like dealing with a rash.

Meta has not provided a comment on this issue, leaving users questioning the privacy and security of their searches on the platform. The public nature of these searches raises concerns about the inadvertent disclosure of personal information and the potential consequences of sharing sensitive queries with a wide audience.