Communicating Differential Privacy Guarantees to Data Subjects
summary
Compares explanations for differential privacy across two studies: a label format that lists which recipients can see raw data under local vs central deployments (local blocks all flows, central only blocks device-to-curator raw data) tested with >700 respondents and improves understanding and trust when augmented with a short process description; and an odds-based visual combining text and icon arrays tested with >900 respondents that increases risk comprehension, perceived sufficiency of information, and willingness to share data while showing sensitivity to different epsilon values.
AI-written summary · about
chapters (8)
- 0:00Introduction to communicating differential privacy
- 1:10DP implementation details
- 2:31Gaps in DP explanations
- 3:55Deployment model with privacy labels
- 5:38Visualizing epsilon odds-based
- 8:26Key findings transparency
- 9:49Lessons for other privacy-enhancing technologies
- 10:10Q&A
related talks
06mLessons Learned in Communicating Differential Privacy to Data Subjects
09mReplication: The Effect of Differential Privacy Communication on German Users'...
20mShare data without revealing personal information
42mDifferential Privacy
11mMy $5MM Differential Privacy Visualizations
58mLarge-Scale Private Learning, Part III
01h 00mPANEL: Statistical Theory, Privacy and Data Analysis
01h 05mMarkets and Incentives for Transacting on (Personal) Information, Part III