Towards Building a Responsible Data Economy
summary
Argues that secure enclaves, cryptographic MPC, differential privacy and federated learning combine with blockchains to make data usable without copying and to preserve rights; demonstrates extraction of credit cards and SSNs from an Enron-trained language model and similar leakage from GPT-2, and shows differentially private training reduces that leakage while retaining utility. Presents Keystone (RISC‑V TEE) as an open source, near‑native enclave implementation and describes Oasis Labs’ data asset/tokenization plus blockchain logging for genomic data use.
AI-written summary · about
chapters (7)
- 0:00Introduction to talk
- 2:29Responsible data economy goals
- 5:18Technical solutions for data economy
- 13:05Secure computing technology
- 16:16Decentralized data science
- 19:44Incentives and legal framework
- 21:08Future of responsible data economy
related talks
27mCreate Value and Insight Through Privacy-preserving Data Collaboration [...]
15mKeynote on Building a Network for Privacy-Preserving DeFi
01h 20mUnlocking societal benefits of AI with privacy protective technologies.
20mDeep Dive into Privacy Enhancing Technologies
44mIf only I owned my data: Architecting decentralized data
01h 24mPseudonymity, Automation, and Responsibility on Public, Permissionless Blockchains under the GDPR
14mToward Provably Private Insights into AI Use
01h 19mWhen Data Can’t Move but Insight Must