Meta will use off-site activity to personalize feeds and AI responses

Meta said it will use data from ad partners, including website activity gathered via the Meta Pixel, to personalize not just targeted ads but user feeds and the responses of its AI chatbot. The company frames this as advancing its goal of 'personal superintelligence' — assistants that know enough about a user's interests and behavior to be deeply tailored. Meta clarified it is not collecting new categories of data and that users retain control via privacy settings.

The move extends ad-targeting infrastructure into AI personalization, a logical but contentious step: the same cross-site behavioral signals that power Meta's ad business would now shape what its AI tells you. That raises familiar privacy and manipulation concerns, amplified because conversational AI responses feel more authoritative and personal than ranked feed posts.

The announcement fits Meta's broader 2026 repositioning — the closed 'Avocado' models, the Scale AI investment, Muse Spark on smart glasses, and the India data center — all pointing toward a more controlled, deeply integrated, data-rich AI stack. Privacy advocates will scrutinize the 'not collecting new data' claim, since the change is about new uses of existing data rather than new collection, and regulators (especially in the EU, where Apple is withholding Siri AI features) may take interest. For users, the practical question is whether more personalized AI responses feel helpful or invasive — and whether the opt-out controls are meaningful in practice. The framing around 'personal superintelligence' signals Meta sees behavioral data as a core differentiator versus rivals lacking comparable consumer-behavior datasets.