Facebook privacy has been a concern for users for some time. The social media giant is notorious for collecting, storing, and processing enormous amounts of user data for highly personalized ads. To address these concerns, Facebook is now introducing new Privacy-Enhancing Technologies (PETs) for limiting the amount of data processed for targeted advertising.
The personalization of Facebook Ads as well as measuring traffic will still be functional features. The newly introduce PETs will modify those features to achieve the aforementioned privacy objective.
Multi-Party Computation (MPC)
MPC is used to report the outcomes of ad campaigns to multiple parties. While MPC will process data such as the online activity of users, the data will have end-to-end encryption. This means that the data being processed cannot be directly accessed by advertisers. The key feature of MPC is the involvement of multiple parties in the analysis of an ad campaign’s performance. One party will have the data about users who saw the ad and while another will have information about users who made the purchase.
The MPC technology is currently in testing and will be publicly available from next year.
On-Device Learning
On-Device Learning will have user data stored and processed on the user’s device itself instead of a remote cloud server. Data related to actions taken by the user will be stored on the device and will be used to gather personalization insights. The data will include patterns of clicking ads as well as items purchased.
Differential Privacy for Facebook
In order to protect the identity of users when committing an action, this technology will alter data sets to provide slightly inaccurate information to advertisers. This will be done by adding or subtracting a random number. So if 100 people purchase a product by clicking in an ad, Facebook will display a number like 95, 98, or 102 instead. This will prevent advertisers from uncovering the exact identity of users purchasing the products.
For more updates, be with Markedium.
