Meta's AI 'Instants' Feature Sparks Privacy Backlash

- The "Instants" feature is a rebranding of a previously tested feature called “Shots,” designed for ephemeral photo sharing similar to Snapchat's core functionality. It allows mutual followers to exchange photos that disappear after 24 hours and cannot be edited, representing another attempt by Meta to compete directly with Snapchat. - User interactions with Meta's AI features, including "Instants" and chatbots, are leveraged to personalize recommendation systems and ad targeting across its platforms. For instance, a conversation with Meta AI about a vacation destination could directly influence the travel-related Reels and ads a user sees on Facebook and Instagram. - Meta's AI terms allow the company to analyze user-uploaded images and facial features to train its generative models. This has amplified privacy concerns, particularly following reports that human contractors have accessed sensitive, raw user data from AI chats—including names, emails, and explicit photos—to improve the AI. - A significant privacy flaw in the standalone Meta AI app led to users unintentionally broadcasting sensitive conversations on a public "Discover" feed. Users seeking advice on medical, legal, and personal matters were unaware their queries were being publicly posted, a situation one security expert described as a “privacy nightmare.” - The introduction of AI-generated accounts and interactions is intended to increase user engagement and retention. However, critics raise concerns about the difficulty of distinguishing between authentic and synthetic engagement, which can pollute the data used to train and evaluate the performance of large-scale recommendation systems. - This is not an isolated incident; Meta has a documented history of privacy violations. Researchers recently discovered the company using a tracking pixel that exploited a loophole in Android's security sandbox to siphon web browsing data from users, even if they used a VPN or cleared their cookies. - The backlash comes amid broader regulatory pressure, with bodies like the European Data Protection Board (EDPB) scrutinizing Meta's use of public user data for training its AI models without explicit opt-in consent. This follows previous major privacy-related penalties, including a €1.2 billion fine for GDPR violations and a $725 million settlement for the Cambridge Analytica data scandal. - Internally, Meta is also reportedly considering adding facial recognition to its Ray-Ban smart glasses. Leaked documents suggest the feature, internally dubbed "Name Tag," would use AI to identify people, raising significant concerns about the potential for real-world, instant doxxing and a further erosion of privacy.