Meta announced on Wednesday that data collected from user interactions with its AI products will soon be used to sell targeted ads across its social media platforms.

The company will update its privacy policy by December 16 to reflect the change, and will notify users in the coming days. The new policy applies globally, except for users in South Korea, the United Kingdom, and the European Union, where privacy laws prevent this type of data collection.

Meta’s core business has long relied on building detailed profiles of Facebook and Instagram users to sell hyper-targeted ads. The company offers advertisers a way to reach specific demographics and user groups. Now, Meta will also use data from conversations with its AI chatbot to build out those profiles, giving it another powerful signal to target its ads.

The social media giant already has lots of information about its users, but Meta AI has created a rich new stream of information. The company says more than a billion people chat with Meta AI every month, and it’s common for users to hold long, detailed conversations with the AI chatbot. So far, Meta has largely given away its AI products for free, but now the company can improve its valuable ad products based on the data it collects.

If a user chats with Meta AI about hiking, for example, the company may show ads for hiking gear. However, Meta spokesperson Emil Vazquez tells TechCrunch that the privacy update is broader than just Meta AI, and applies to the company’s other AI offerings.

That means Meta may use data from AI features in its Ray-Ban Meta smart glasses — including voice recordings, pictures, and videos analyzed with AI — to further target its ad products. Meta may also use data from its new AI-video feed, Vibes, and its AI image generation product, Imagine.

Conversations with Meta AI will only influence ads on Facebook and Instagram if a user is logged into the same account across products.

There is no way to opt out, according to Meta.

The privacy changes are another reminder that free products from Big Tech companies often come with strings attached. Many tech companies already use AI interactions to train their models. Meta, for instance, trains on voice recordings, photos, and videos analyzed through Meta AI on its smart glasses. Now it will also feed that data into its ad machine.

In a briefing with reporters, Meta privacy policy manager Christy Harris said the company is still in the process of building out systems that will use AI interactions to improve its ad products. However, the company says user conversations with AI around sensitive topics — including religious views, sexual orientation, political views, health, racial or ethnic origin, philosophical beliefs, or trade union membership — will not be used to show them ads.

Tech companies are starting to test out ways to monetize AI products, most of which are free today. On Monday, OpenAI unveiled a way to purchase products in ChatGPT, where the company will take a cut of transactions completed in the app. Earlier this year, Google revealed plans for how it would introduce ads into its AI-powered search product, called AI Mode.

Meta says the company has “no plans imminently” to put ads in its AI products, though CEO Mark Zuckerberg has suggested they may be coming in the future.