Meta to use AI conversations for ad personalisation (about.fb.com)

Meta announced it will start using people’s interactions with Meta AI — both text and voice chats — as a signal to personalize content and ads across its apps, beginning Dec. 16, 2025 (users will be notified by email/notifications several weeks beforehand). Meta positions these AI conversations like other engagement signals (likes, posts, follows): if you ask about hiking, for example, you may see more hiking reels, groups, or related ads. The change applies across Meta Company Products when accounts are linked in the Accounts Center; interactions on unlinked accounts (e.g., WhatsApp not added) won’t be shared across other products. For the AI/ML community this expands the behavioral data feeding recommender systems and ad models — potentially improving personalization but raising privacy and policy questions. Meta says it won’t use conversations about protected or sensitive attributes (religion, sexual orientation, political views, health, ethnicity, trade-union membership) for ad targeting, and voice use requires explicit microphone permission and shows an indicator when active. Users can still adjust Ads Preferences and feed controls. Technical implications include richer contextual signals for training and real-time ranking, broader cross-product feature engineering via Accounts Center links, and increased regulatory and ethical scrutiny around inferred preferences from conversational data.