Dynamic De-Identification for Privacy-Safe AI User Input
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to provide dynamic and controlled data anonymity and privacy, especially in decentralized networks, leading to tensions between maximizing data value and respecting individual privacy rights, and they are vulnerable to re-identification and data misuse.
Innovation Solution
Implementing dynamically changing de-identifiers (DDIDs) associated with data subjects for temporally unique periods, enabling controlled and flexible anonymity, allowing data sharing only when authorized and within specified contexts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is shared to maximize data value for AI/ML analytics, then productivity and data utility are improved, but privacy and security of data subjects deteriorate due to re-identification risks
Solution Approach 1:
The patent segments data into identified and unidentified portions, allowing AI/ML models to access utility-rich identified data while keeping sensitive personal information separated and protected. This segmentation enables simultaneous achievement of data productivity and privacy security by ensuring that even if identified data is accessed, the core personally identifiable information remains isolated and secure.
Solution Approach 2:
The patent introduces an intermediary data structure (temporally unique data representation or TDR) that acts as a mediator between data subjects and AI/ML systems. The TDR contains identified data for analytics while excluding sensitive personal information, thereby enabling data utility without compromising privacy security. This intermediary structure resolves the contradiction by providing a controlled interface that delivers value while maintaining protection.
2Reliability
If static de-identification methods are used to protect privacy, then privacy security is improved, but data accuracy and personalization capabilities deteriorate
Solution Approach 1:
The patent replaces static de-identification with dynamic de-identification where the level and type of identification change over time based on temporal parameters. Data subjects can transition between identified and unidentified states, allowing them to provide accurate information when needed while maintaining privacy protection at other times. This dynamic approach resolves the contradiction by enabling both privacy security and data accuracy through time-based flexibility rather than permanent static states.
3Reliability
If data subjects maintain continuous anonymity to protect privacy, then privacy security is improved, but data personalization and targeted analytics capabilities deteriorate
Solution Approach 1:
The patent implements periodic transitions between identified and unidentified states for data subjects. During identified periods, data subjects can provide personalization information for targeted analytics and personalization. During unidentified periods, privacy security is enhanced. This periodic cycling resolves the contradiction by ensuring that personalization information is available when needed while maintaining privacy protection during other times, rather than requiring continuous anonymity.
4Measurement precision
If identified data is collected to improve data accuracy for AI models, then measurement precision is improved, but vulnerability to re-identification and data misuse increases
Solution Approach 1:
The patent applies preliminary filtering and selection to data before it is made available to AI/ML systems. The TDR structure pre-excludes sensitive personal information from identified data portions, so that even accurate data is provided without the harmful re-identification risk factors. This preliminary action resolves the contradiction by ensuring data accuracy is maintained while dangerous personal information is removed in advance, preventing re-identification vulnerabilities before they can occur.
Data Source
AI summary
Various systems, computer-readable media, and computer-implemented methods of providing improved data privacy, anonymity and security by enabling subjects to which data pertains to remain “dynamically anonymous,” i.e., anonymous for as long as is desired—and to the extent that is desired—are disclosed herein. Embodiments include systems that create, access, use, store and/or erase data with increased privacy, anonymity, and security—thereby facilitating the availability of more qualified and accurate information. When personal data is authorized by data subjects to be shared with third parties, embodiments described herein may facilitate the sharing of information in a dynamically-controlled manner that also enables the delivery of temporally-, geographically-, and/or purpose-limited information to the receiving party. In one example, the disclosed techniques may be used to functionally separate geospatial information, such that it remains “dynamically anonymous,” i.e., anonymous for as long as is desired—and to the extent or degree that is desired.