Meta to Ingest AI Conversations for Ad Curation

Meta Platforms will begin integrating users’ interactions with its generative AI tools into content and ad recommendations across Facebook, Instagram, and linked services from 16 December 2025. Users will receive in-app notifications from 7 October informing them of the change, but those who engage with Meta AI will have no option to opt out — though the update will initially omit the United Kingdom, the European Union and South Korea, owing to regulatory constraints.

The shift marks a pivotal move in Meta’s data strategy: voice and text conversations with its AI systems will now join traditional signals — such as likes, follows and shares — in shaping personalised feeds and advertisements. The company has emphasised that chats on sensitive issues like politics, religion, health, sexuality, race or trade union membership will be excluded from targeting decisions.

Christy Harris, Meta’s privacy policy manager, said users’ AI interactions will “simply be another piece of the input” informing content and ad personalization. The approach aims to deepen relevance, enabling, for instance, a user who discusses hiking with Meta AI to see greater content related to trails or ads for outdoor gear.

Meta reports that over one billion people use Meta AI every month. The company intends to harness that engagement as a new signalling layer, effectively enhancing its predictive advertising models. In a blog post, Meta reiterated that users may adjust their broader ad preferences, but once the new system is live, there will be no mechanism to opt out of AI-based targeting for those who use Meta AI.

Rights groups and privacy advocates are already voicing alarm at the potential for intrusive profiling. Critics point to the inherent ambiguity in differentiating “sensitive” from “non-sensitive” topics: what one user considers innocuous may be interpreted as personal data. Moreover, a recent study demonstrated that AI-mediated ad systems can reconstruct demographic traits or behavioural profiles from ostensibly anonymous ad exposures.

Meta is not alone in exploring conversational data monetization. Google, Amazon and OpenAI have all explored channels whereby AI interactions inform recommendations, commerce or ad delivery. Analysts see this as part of an emerging industry pattern: free AI services anchored by surveillance-style data harvesting.

Underlying the controversy is a tension between personalisation and user autonomy. Independent audits have revealed that ad controls and transparency tools often fall short, failing to meaningfully empower users to shape the content they see under newer AI-driven targeting systems.

Meta has not disclosed the full technical details of how it will filter sensitive topic conversations, or how much weight AI interactions will carry relative to existing signals. Its algorithms may treat conversational data as a higher signal intensity, given the presumed depth of user interest expressed in dialogue versus casual interaction.

Internal reports and third-party assessments suggest that human reviewers have accessed parts of AI chat logs — including personally identifiable information and intimate content — during system training and oversight phases. Such exposure intensifies apprehensions about automated decision systems built on deeply personal data.

Regulators in Europe and the UK may delay enforcement of the new model, giving authorities time to evaluate compliance with data protection frameworks such as the General Data Protection Regulation. Meta’s exclusion of those jurisdictions at launch reflects the heightened scrutiny it expects.