Meta is changing the prompt suggestions generated by its AI chatbot. The update follows a viral video showing the chatbot asking invasive questions regarding a woman and her young daughters. Futurism initially reported the incident.
Meta spokesperson Dina El-Kassaby stated that the company missed the mark. She added that the feature never should have prompted the individual with questions like that.
Instagram user Kalie Robins posted a video last week detailing the event. She explained that Meta AI presented an invasive suggestion after she cross-posted a video clip to Facebook.
The prompt asked who the child passenger was beneath a video showing Robins and her child singing inside a car. Robins selected the prompt, leading Meta AI to piece together information about her daughters from her prior posts and posts by her relatives.
The system suggested additional prompts regarding the ages of the children and the family residence location. Robins stated that Meta AI also displayed photos of her daughters, including one she claims to have deleted years earlier.
Meta integrates its AI assistant into Facebook, Instagram, WhatsApp, and Messenger. The assistant can answer questions, draft posts, and generate images.
The company previously pulled a feature in July that allowed people to create AI deepfakes of other Instagram users. That decision followed fierce public backlash.
El-Kassaby confirmed that Meta has now fixed the issue causing Meta AI to suggest prompts related to personal topics. She explained that the company created the feature to help people get more information about topics or posts that interest them.
When users click a prompt and use this feature, the generated response relies on information the user already can access. El-Kassaby noted that in this specific case, Meta AI may have pulled from videos where Robins named her children.
Meta is currently deploying the prompt changes to address the issue.
