If a specific ad is being targeted at only one single person, then anyone who can learn that that specific ad appeared on a particular website could immediately infer that the one person it was targeting visited that website. To avoid this sort of privacy risk, the Protected Audience API will only show ads with a large enough audience. Requiring an audience of size at least k, where k is some minimum number of people, is sometimes called "k-anonymity". You can learn more about k-anonymity thresholds in the developer documentation.
In addition to k-anonymity, Protected Audience incorporates other privacy-preserving measures:
- Secure processing: Ad auctions and data processing happen directly on the user's device, reducing the amount of data shared with third parties. Alternatively, the Bidding and Auction services (B&A) proposal outlines a way to allow Protected Audience computation to take place on cloud servers in a trusted execution environment (TEE), rather than running locally on a user's device. Moving computation to servers can help optimize the Protected Audience auction by freeing up computational cycles and network bandwidth for a device.
- Limited data sharing: The API is designed to limit the amount of data shared between different parties, reducing the risk of unauthorized access or leakage.
- Fenced frames: Ads can be displayed within fenced frames, which isolate them from the rest of the web page and protect against potential tracking or data leakage.
Learn more about how Protected Audience helps safeguard user privacy in the developer documentation.