Personal Data Pseudonymization Versus Anonymization In The Age Of AI & Big Data
summary
Claims that naive identifier removal fails, backed by numbers: 15 attributes → 99.8% re-identification and six attributes already give ~50% chance. Concrete incidents include a 200,000-student dataset where small cohorts let parents infer individual scores and a staging site left linked to production that leaked orders. Practical controls discussed are not vague promises but tools and patterns: encryption + separated keys, SMPC, federated learning, TEEs, zero‑knowledge proofs, synthetic data and differential privacy, plus dataflow mapping and DPIAs to test whether anonymization actually works.
AI-written summary · about
chapters (6)
- 0:00Introduction and speaker intro
- 1:49Definition of personal data
- 10:44Big data and personal data challenges
- 17:10Anonymization vs pseudonymization
- 22:31Privacy enhancing technologies
- 27:26Compliance and next steps
related talks
32mThe Dangerous Data Anonymization - Keynote
33mPractical Data Privacy
01h 17mTo be or not to be (anonymous)? Anonymyity in the age of big and open data.
10mHealth Data Anonymization and Privacy
09mPrivacy Analytics and De-Identification
50mPrivacy in Practice with Smart Pseudonymization: Lessons from the Belgian Public Sector
50mData Privacy Concerns - A Risk Based Approach to De-Identification
15mHow We Can Save Anonymization