Sharing without Sharing: the Case for Data Clean Rooms Powered by Confidential Computing
summary
Claims that a Decentriq data clean room lets multiple organizations run joins and ML on sensitive records without exposing raw inputs, using hardware confidential Computing (trusted execution environments) plus aggregation, differential privacy and synthetic-data tools. Platform supports SQL and Python, scales to large workloads with about a 15–20% performance hit versus non‑protected runs, and runs live cases: hospital+pharmacy+pharma study (receiving Belgian DPA approval) and a bank+publisher marketing deployment. The system is built and operating in production.
AI-written summary · about
chapters (5)
- 0:00Start
- 1:35Sharing without Sharing
- 4:09Input Privacy
- 9:07Confidential Computing
- 12:41The Clean Room Concept
related talks
15mA Large-Scale Data Clean Room Case Study in Japan
39mData Privacy Engineering to power Responsible Enterprise AI
45mConfidential Computing: Protecting Customer Data in the Cloud
23mConfidential computing: from niche to mainstream
19mPreserving privacy on data collaboration with AWS Clean Rooms
13mProvably Secure AI for Health Data – Powered by Confidential Computing
36mCTO Panel: The Impact of Confidential Computing with Intel, Azure, AMD & NVIDIA
17mTowards the medicine of the future with confidential computing