Meta changes Meta AI suggestions after invasive questions in a viral video

In short: Meta says it has changed the suggested prompts in Meta AI after the chatbot surfaced personal questions about a user’s children.

What happened

Meta told The Verge it is making changes to the suggested prompts that appear in its Meta AI chatbot inside apps like Facebook and Instagram. The company said the feature “missed the mark” and “never should have” suggested questions like the ones shown in a viral video.

The video was posted by Instagram user Kalie Robins. She said that after she shared a clip of herself and her child singing in a car, Meta AI showed a suggested prompt under the video that asked, “Who is the child passenger?” Robins said the prompt felt intrusive because it pushed her to identify a child.

Robins said that when she clicked the prompt, Meta AI appeared to pull together details about her daughters from older posts and from posts made by relatives. She said it suggested more prompts, including questions about her children’s ages and where her family lives. She also said Meta AI displayed photos of her daughters, including one she believes she deleted years ago.

Meta spokesperson Dina El-Kassaby said the company has fixed the issue that caused prompts related to personal topics to appear. Meta also said that when someone clicks a prompt, the AI’s response is based on information the user already has access to, and that in this case it may have pulled from videos where Robins named her children.

Why it matters

Suggested prompts are like a row of “quick questions” meant to help people interact with a chatbot. If those questions push users toward sharing private details, especially about children, it can create real safety and privacy concerns and make people less willing to post family content.

Source: The Verge AI