Meta’s AI Prompt Scare Rekindles Parents’ Fears Over Children Online

Meta’s AI Prompt Scare Rekindles Parents’ Fears Over Children Online
The alarm began last week when Utah mother Kalie Robins cross-posted a video of herself singing with her child in a car from Instagram to Facebook. Beneath it, Meta AI offered a question that stopped her cold: “Who is the child passenger?”

Robins said that after she clicked, the chatbot appeared to assemble details about her daughters from her own older posts and material posted by relatives. It suggested further questions about the children’s ages and the family’s home, and displayed photos of the girls — including one Robins said she had deleted years earlier.

Meta’s response was an admission, not a denial. The company said the answers generated after a user selected a prompt relied on information already available to that user, but conceded the prompts themselves should never have been offered. “We missed the mark,” spokesperson Dina El-Kassaby said, adding that the underlying issue had been fixed.

For Robins, that distinction offers little comfort. She said she had avoided obvious identifiers: no school names, teacher details, street signs or landmark-heavy photos. “I thought I was being so careful and ultimately that didn’t matter,” she said, because information had come from other people’s posts.

The incident landed as another Instagram-driven shift was already under way. Family photographers have begun announcing that they will no longer publish children’s faces, even with parental permission, citing fears that publicly available images could be scraped into systems used to generate abusive AI material.

Kentucky photographer Hannah Haggard said a recent legal ruling involving AI-generated child sexual-abuse material solidified a policy she had already been considering. “Then once the ruling happened, it was confirmation that that’s what needed to happen,” she said.

The immediate Meta failure and the broader panic are not identical: the company says its bot accessed existing platform information, while photographers worry about what AI models may learn from images at scale. But both point to the same hard lesson for families: online privacy is no longer defined solely by what a parent chooses to post.

Continue reading https://foxvector.com/stories/01a0937f-2171-04f9-708a-0108db2bb78f