Meta says it has corrected an AI feature that generated personal questions about Instagram users after a mother demonstrated how suggested prompts could rapidly assemble information about her family and minor children from years of social media posts.

By yourNEWS Media Newsroom.

Meta says it has fixed an artificial intelligence feature that suggested personal questions about Instagram users and their children after a parenting creator’s viral demonstration showed the system assembling family information from multiple social media posts.

The company told CNET that Meta AI will no longer generate what it described as inappropriate personal questions beneath posts following complaints from Instagram creator Kalie Robins, known online as kontheinside.

Robins raised the issue after Meta AI generated suggested prompts beneath a video showing her singing inside a car while her daughter sat in the backseat.

Among the questions displayed were “Who’s the child passenger?” and “Where does Kalie Robins live?”

Robins said selecting the prompts caused Meta AI to search information already available across her social media history and quickly assemble details about her family, including her location and information concerning her children.

The system did not limit itself to her own account, according to Robins. She said it also gathered material from other profiles, including her mother’s account, and surfaced information connected to a photograph she believed she had deleted years earlier.

“I did not understand posting separate innocent pieces of our lives over years could mean an AI could sit there and assemble all those breadcrumbs into information about my minor children in seconds,” Robins said in her viral Instagram video.

Her post attracted thousands of comments and more than 308,000 likes, bringing wider attention to how artificial intelligence can combine publicly available fragments of personal information far more rapidly than an individual browsing someone’s account manually.

The episode also prompted some parents to question the long-term privacy implications of sharing photographs and information about children online, with Business Insider reporting that parents were encouraging others to reconsider posting images of their children on social media.

Meta said the suggested prompts were not functioning as intended.

A company spokesperson told CNET by email that the problem had been “fixed” and said Meta AI would no longer suggest inappropriate personal questions on posts.

According to Meta, the feature is intended to help users obtain additional information using material they are already permitted to access, rather than reveal private information unavailable to them.

The company also disputed one portion of Robins’ account involving the photograph she believed had been deleted years earlier.

Meta said the image had actually been removed shortly before Robins made her video. According to the company, a temporary bug caused the recently deleted photograph to remain visible for a short period afterward.

Robins challenged that explanation and said the larger concern was not how long one particular photograph remained accessible.

In a Sept. 10 follow-up post, she called on Meta to explain what specifically had gone wrong and what changes had been made to prevent similar incidents involving children.

“What exactly was fixed? What safeguard failed in the first place? And what protections are now in place when Meta AI encounters information involving a child?” Robins asked.

She made the same distinction in an email to CNET.

“The issue was never just one photo, it was the AI compiling information about my kids from multiple sources,” Robins said.

She also said Meta had neither contacted her directly nor apologized.

The episode highlights a privacy issue that differs from the traditional problem of whether an individual piece of information has been posted publicly.

A user may intentionally share separate photographs, family references, locations and other details over a period of years without considering how quickly an AI system could combine those fragments into a detailed profile.

Robins said that was precisely what disturbed her: Information that appeared harmless when viewed separately could be collected and connected within seconds when processed by an automated system.

Her complaint also places particular focus on minors, who may have extensive digital histories created by parents before they are old enough to make their own decisions about what information should appear online.

Meta maintains that its AI feature is designed to work only with information users can already access and says it has corrected the inappropriate prompt suggestions that brought the issue to public attention.

Robins, however, continues to press the company for greater transparency about why the system generated questions concerning her child and location in the first place and what technical safeguards now prevent a recurrence.

The controversy has consequently moved beyond a single malfunctioning prompt. It raises a broader question about how AI systems can transform scattered pieces of ordinary social media activity into consolidated personal information — and what additional protections may be necessary when that information concerns children.

Original article