Meta says it’s changing AI suggestions after posing invasive personal questions

Meta has announced modifications to its artificial intelligence assistant's suggestion feature following public backlash over invasive automated prompts.

The decision came after an Instagram creator shared a viral video demonstrating how the integrated AI chatbot questioned her about her young daughters' identities beneath a cross-posted Facebook clip. After she interacted with the initial query, the software began compiling highly personal details, such as her children's ages and geographic location, by cross-referencing past social media posts from both her account and those of her relatives. The user also reported that the system retrieved and displayed a photograph she had deleted years prior, raising significant privacy concerns.

In response to the incident, Meta acknowledged the error, admitting that the algorithm missed the mark and should not have targeted the individual with such intimate queries. The technology giant explained that the system was originally designed to help users discover more context about subjects of interest by drawing on accessible network data. However, the company has since implemented a fix to prevent the assistant from generating automated prompts related to highly sensitive, personal topics. This controversy follows another recent backtrack by Meta in July, when the company disabled an interactive feature that let individuals generate AI-driven deepfakes of other users on its platforms.