Preserving privacy on data collaboration with AWS Clean Rooms
summary
Explains how AWS Clean Rooms uses AWS Glue and S3 to run SQL over joined datasets while enforcing rules like aggregation with minimum threshold and blind match on a designated join column so collaborators can match customers without revealing raw identifiers. Describes using client-side encryption (C3R) to keep plaintext locally, and configuring a privacy budget and noise for differential privacy; cites the 1997 Governor re‑identification example and points to NIST guidance for output privacy.
AI-written summary · about
chapters (15)
- 0:00Introduction to privacy challenges
- 2:33Handling PII in data
- 3:37Usability vs privacy trade-off
- 5:00AWS Clean Rooms overview
- 6:04Aggregation rules and thresholds
- 7:29Blind matching with list rules
- 8:14Cryptographic computing in Clean Rooms
- 8:56Output privacy risks
- 10:42Differential privacy concepts
- 12:49Setting up Clean Room collaboration
- 13:52Querying data in Clean Rooms
- 15:16Configuring privacy budget and noise
- 16:18Key takeaways and recommendations
- 17:46Re-identification and NIST guidelines
- 18:30Q&A and resources
related talks
39mData Privacy Engineering to power Responsible Enterprise AI
30mSharing without Sharing: the Case for Data Clean Rooms Powered by Confidential Computing
56mUnderstanding and achieving a modern data architecture
28mTurn Data Into Insights
49mBuild modern data architecture and data mesh patterns on AWS
16mAutomated Data Analytics
28mDeep Dive Into AWS Lake Formation
01h 59mDeep Dive Workshop – Data Security and Data Protection