Differential privacy policy - AWS Clean Rooms

Differential privacy policy

The differential privacy policy controls how many aggregation functions the member who can query is allowed to run in a collaboration. The Privacy budget defines a common, finite resource that is applied all tables in a collaboration. The Noise added per query governs the rate at which the privacy budget is depleted.

A differential privacy policy is required to make your differential privacy protected tables available for querying. This is a one-time step in a collaboration and includes two inputs:

  • Privacy budget – Quantified in terms of epsilon, the privacy budget controls the level of privacy protection. It is a common, finite resource that is applied for all of your tables protected with differential privacy in the collaboration, because the goal is to preserve the privacy of your users whose information can be present in multiple tables.

    The Privacy budget is consumed every time a query is run on your tables. When the privacy budget is fully exhausted, the collaboration member who can query can't run additional queries until it is increased or refreshed. By setting a larger privacy budget, the member who can receive results can reduce their uncertainty about individuals within the data. Choose a privacy budget that balances your collaboration requirements against your privacy needs and after consulting with business decision-makers.

    You can select Refresh privacy budget monthly to automatically create a new privacy budget each calendar month, if you plan to regularly bring new data into the collaboration. Choosing this option allows arbitrary amounts of information to be revealed about rows of the data when repeatedly queried across refreshes. Avoid choosing this if the same rows will be repeatedly queried between privacy budget refreshes.

  • Noise added per query is measured in terms of the number of users whose contributions you want to obscure. This value governs the rate at which the privacy budget is depleted. A larger noise value reduces the rate at which the privacy budget is depleted, and therefore allows more queries to be run on your data. However, this should be balanced against releasing less accurate data insights. Consider the desired accuracy for collaboration insights when setting this value.

You can use the default differential privacy policy to quickly complete the setup or customize your differential privacy policy as per your use case. AWS Clean Rooms Differential Privacy provides intuitive controls to configure the policy. AWS Clean Rooms Differential Privacy lets you preview the utility in terms of the number of aggregations possible across all queries on your data and estimate how many queries can be run in a data collaboration.

You can use the interactive examples to understand how different values of Privacy budget and Noise added per query would impact the results for different types of SQL queries. In general, you need to balance your privacy needs against the number of queries you want to permit and the accuracy of those queries. A smaller Privacy budget or larger Noise added per query can better protect user privacy, but provides less meaningful insights to your collaboration partners.

If you increase the Privacy budget while keeping the Noise added per query parameter the same, the member who can query can run more aggregations on your tables in the collaboration. You can increase the Privacy budget any time during the collaboration. If you decrease the Privacy budget while keeping the Noise added per query parameter the same, the member who can query can run fewer aggregations. You can't decrease the Privacy budget after the member who can query has started analyzing your data.

If you increase the Noise added per query while keeping the Privacy budget input the same, the member who can query can run more aggregations on your tables in the collaboration. If you decrease the Noise added per query while keeping the Privacy budget input the same, the member who can query can run fewer aggregations. You can increase or decrease the Noise added per query any time during the collaboration.

The differential privacy policy is managed by the privacy budget template API actions.