Meta modifies Meta AI after alert on family data collection
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
After a viral video from a mother, Meta claims to have fixed a Meta AI feature that suggested questions that could aggregate personal information, including about children. However, the company still maintains suggestions based on accessible content. Disagreement persists over access to a supposedly deleted photo, amid significant pressure regarding the protection of minors.
Meta Acts Under Intense Pressure Related to Children and Privacy
Meta has faced multi-billion dollar regulations and court rulings this year concerning the impact of its technologies on children, particularly regarding addiction and vulnerability to exploitation. Additionally, its smart glasses have come under criticism, especially for the facial recognition technology in development. Meanwhile, Meta is beginning to roll out its Muse AI agents, designed to offer more personalized AI experiences through personal assistants.
What Meta Has Changed and What AI Continues to Do
Meta states it has fixed an issue that allowed Meta AI to suggest inappropriate personal questions about users' posts. The company acknowledges that the feature did not function as intended and clarifies that the tool is designed to help users obtain information from data they already have access to. Despite the removal of intrusive prompts, Meta AI can still provide suggestions based on accessible information, such as public texts, images from the web, or previous posts. Meta presents this correction as a response to the controversy sparked by a viral video.
The September 2 Video and Reported Intrusive Prompts
On September 2, blogger Kalie Robins, known as kontheinside, posted a video on Instagram discussing an old post where she sang in the car with her daughter in the back seat. She claims that Meta AI suggested questions like "Who is the child passenger?" and "Where does Kalie Robins live?" According to Robins, by clicking on these automatic questions, the AI extracted information from her past posts to reconstruct private details, including her location, and used other profiles, such as her mother's. She also reports that the AI accessed data from a photo she claims to have deleted years ago. Robins expressed her surprise at the AI's ability to quickly assemble information about her minor children from fragments of life shared online. Her video went viral, garnering thousands of comments and over 308,000 likes.
Meta's Denial Regarding the Photo and Unanswered Questions
Meta denies having accessed a photo deleted several years ago, specifying that the image of Kalie Robins' children was deleted just before the video was posted and that a temporary malfunction made it accessible for a brief period. In a message posted on September 10, Robins expressed her satisfaction that the issue had been acknowledged while demanding clarification on the corrections made, the initial protection flaw, and the measures now applied when Meta AI processes data related to a child. She emphasizes that the concern goes beyond the case of the mentioned photo and notes that she has not been contacted directly nor received an apology from Meta.
Chain Reactions Among Parents and Alerts from a Creator
The media coverage of the incident, particularly on Futurism, has led some parents to recommend no longer posting photos of children online. Cathy Pedrayes, a creator specializing in topics related to scams and dark web trends, reacted by calling the situation concerning.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.