Meta says that beginning December 16 it will use what people tell its AI as a targeting signal for the posts, reels and ads they see across Facebook, Instagram and the company’s other apps.

The company frames the change as an extension of existing personalization: if someone asks Meta AI about hiking, it may infer they like hiking, much as it would from a liked Page or a posted reel, and serve trail content or ads for boots. Meta says it will start notifying people on October 7 by in-product notifications and emails. There is no opt-out, per Meta; users keep access to the same ad-preference and feed controls they had before.

The policy covers voice chats and text exchanges with Meta AI, and Meta says it applies more broadly to its other AI features. Conversations about religion, sexual orientation, politics, health, race or ethnicity, philosophical beliefs or trade union membership are excluded from ad personalization. Users in the EU, the UK and South Korea are not affected. Meta puts monthly Meta AI use above one billion people.

The objections arrived fast. Watchdog researchers noted that conversations are another proprietary stream of targeting data that outside brands cannot audit; one privacy specialist said the plumbing to actually do this is still under construction.