Meta says it’s changing AI suggestions after posing invasive personal questions

AI chatbots now suggest invasive questions after accessing your personal posts, forcing you to re-evaluate what data is publicly visible.

Meta is modifying its AI chatbot's suggestion feature after a user reported invasive questioning about her children. The AI reportedly accessed personal information from past posts to suggest prompts about the user's daughters' ages and location. This occurred after the user, Kalie Robins, posted a video of herself and her child singing in a car. The AI's prompt, "Who is the child passenger?", led to the disclosure of private details, including photos, some of which the user claimed to have deleted years ago. Meta acknowledged the feature "missed the mark" and "never should have prompted the individual with questions like that." The company stated the feature is intended to provide information based on user-accessible data and that the issue causing these prompts has been resolved.

This incident highlights the risks associated with AI's rapid integration into social media platforms, where the aggregation of user data for contextual AI prompts can lead to privacy breaches. Meta's admission of error and subsequent fix indicate a reactive approach to AI safety, potentially indicating that other AI features across various platforms may carry similar unaddressed vulnerabilities. The company's reliance on "user-accessible information" for AI responses raises questions about the ethical boundaries of data scraping, even when technically permissible.

The backlash Meta faced, mirroring previous issues with AI deepfakes, suggests a broader market hesitancy in trusting AI's ability to handle sensitive personal information responsibly. As AI becomes more deeply embedded in communication tools, users will likely become more cautious about the data they share and the AI functionalities they enable, potentially slowing the adoption of AI features that require extensive personal data access.