Lessons from years of building privacy-preserving technologies
summary
After thousands of hours collaborating with researchers, regulators and practitioners, the conclusion is that TEEs plus cryptographic hardening hit the best balance for analytics and AI: cryptographic MPC/FHE remains too slow for big data while a secure SQL analytics engine runs only about 2× slower than plaintext on business query suites. Opaque builds a platform with policies: data, computation, output, layered crypto, familiar APIs and multi‑cluster scaling to enable verifiable, governed collaboration.
AI-written summary · about
chapters (7)
- 0:00Introduction to privacy tech
- 1:45Why privacy-preserving tech matters
- 3:12Data protection in three parts
- 5:39Crypto vs hardware approaches
- 7:25Friction in adoption
- 10:32Scalability challenges
- 11:58Summary and Q&A
related talks
40mConfidential Computing and Privacy-Enhancing Technologies - The Landscape
44mIntroduction to confidential computing – protecting sensitive workloads
34mTeaclave: Making Computation on Privacy-Sensitive Data Safe and Simple
39mData Privacy Engineering to power Responsible Enterprise AI
37mBoF: PETs - Privacy-Enhancing Technologies Compared, Discussed &...
37mDemystifying Privacy Preserving Computing
37mPETs - A Technical Solution To Digital Sovereignty?
19mHow Can We Have Private Access to Intelligence?