AWS Clean Rooms Differential Privacy

AWS Clean Rooms Differential Privacy helps you protect the privacy of your users with a mathematically-backed technique that is implemented with intuitive controls in a few clicks. As a fully managed capability, no prior differential privacy experience is needed to help you prevent the re-identification of your users. AWS Clean Rooms automatically adds a carefully calibrated amount of noise to query results at runtime in order to help protect your individual-level data.

AWS Clean Rooms Differential Privacy supports a wide range of analytical queries and is a good fit for a wide variety of use cases, where a small amount of error in the query results will not compromise the usefulness of your analysis. With it, your partners can generate business-critical insights about advertising campaigns, investment decisions, clinical research, and more, all without requiring any additional setup from your partners.

AWS Clean Rooms Differential Privacy protects against overflow or invalid cast errors that make use of scalar functions or math operator symbols in a malicious manner.

For more information about AWS Clean Rooms Differential Privacy, see the following topics.

Differential privacy

Differential privacy allows only aggregated insights and obfuscates the contribution of any individual's data in those insights. Differential privacy protects the collaboration data from the member who can receive results learning about a specific individual. Without differential privacy, the member who can receive results can attempt to infer individual user data by adding or removing records about an individual and observing the difference in query results.

When differential privacy is turned on, a specified amount of noise is added to the query results to obfuscate the contribution of individual users. If the member who can receive results tries to observe the difference in query results after removing records about an individual from their dataset, the variability in the query result helps prevent the identification of the individual's data. AWS Clean Rooms Differential Privacy uses the SampCert sampler, a proven correct sampler implementation developed by AWS.

How Differential Privacy in AWS Clean Rooms works

The workflow to turn on differential privacy in AWS Clean Rooms requires the following additional steps when completing the workflow for AWS Clean Rooms:

After you complete these steps, the member who can query can start running queries on differential privacy protected data. AWS Clean Rooms returns results that comply with the differential privacy policy. AWS Clean Rooms Differential Privacy tracks the estimated number of remaining queries that you can run, similar to the gas gauge in a car that shows you the car's current fuel level. The number of queries that the member who can query can run is limited by the Privacy budget and Noise added per query parameters that are set in the Differential privacy policy.

Considerations

When using differential privacy in AWS Clean Rooms, consider the following: