BoF: PETs - Privacy-Enhancing Technologies Compared, Discussed &...
summary
Session compares practical PETs and when to use each: Confidential Computing (TEE hardware, near real‑time, typically under ~20% overhead, now available on NVIDIA H100-class GPUs), Fully Homomorphic Encryption (mathematically strong but hundreds–thousands× slower, limited ops), Secure MPC (general computation, network/communication bottleneck), Differential Privacy (fast but lossy, leak budget accumulates). Notes 13 open‑source Confidential Computing projects, concrete uses (cross‑bank fraud analytics, multi‑hospital cancer research, UN policy aggregation) and emphasizes key management and layering multiple PETs for defense in depth.
AI-written summary · about
chapters (13)
- 0:00PETs introduction
- 1:27Privacy-enhancing technologies overview
- 5:26Trusted execution environments
- 6:55Fully homomorphic encryption basics
- 10:06Secure multiparty computation
- 13:39Differential privacy role
- 17:16Performance vs security
- 21:16Data use cases
- 24:47Practical considerations
- 26:34Open source tools
- 30:00Key management and threats
- 34:36Defense in depth
- 35:58Risk model with PETs
related talks
37mPETs - A Technical Solution To Digital Sovereignty?
40mConfidential Computing and Privacy-Enhancing Technologies - The Landscape
01h 17mBridging the Gap with PETs Governments' Opportunity to Lead on Innovation and Privacy
44mIntroduction to confidential computing – protecting sensitive workloads
37mDemystifying Privacy Preserving Computing
22mNext-Gen PeT: the Future of Multi-Party Computation Software
12mLessons from years of building privacy-preserving technologies
47mThe Rise of Confidential Computing, What It Is and What it Means to You