AWS Clean Rooms is a secure collaboration workspace that allows organizations and their partners to analyze collective datasets without revealing the underlying data to each other. Enables the creation of “clean rooms” in minutes, allowing users to analyze shared datasets securely. Facilitates collaboration with thousands of companies already using AWS, without needing to move data out of AWS or into another platform. Data remains in its original location within AWS, and AWS Clean Rooms applies built-in analysis rules to maintain control over data during collaboration.

Data Access Controls and Audit Support:

• Analysis Rules: Restrict SQL queries and set output constraints to maintain data privacy and control.
• Cryptographic Computing: Keeps data encrypted even as queries are processed, adhering to stringent data handling policies.
• Query Logs: Logs all queries for review, helping support audits and ensuring compliance.
• Differential Privacy: Protects against user-identification attempts using mathematically-backed techniques. This fully managed capability helps protect user privacy with intuitive controls.
• Machine Learning (ML) Integration: Allows two parties to identify similar users in their data without sharing raw data. A lookalike model can be created and configured using training data, and a lookalike segment can be generated to resemble the training data

How AWS Clean Rooms Works:

• Creating Collaborations:
  • Users create a collaboration and invite AWS accounts to join or join an existing collaboration by creating a membership.
  • Data resources are linked for the use case, including configured tables for event data, models for ML modeling, or ID namespaces for entity resolution.
• Analysis Templates:
  • Users can create or approve analysis templates to pre-agree on the exact queries allowed in a collaboration.
• Data Analysis:
  • Joint data is analyzed by running SQL queries on the configured tables, performing entity resolution using ID mapping tables, or using ML modeling to generate lookalike audience segments.