S0111RS3-T06-S0111-Z · Full risk code

Identity Inference by Correlating Fragmented Data

零散資訊關聯推斷身分

Operation & Monitoring
Risk Description

When a dataset the organization considered de-identified is cross-correlated by a model with other public information, allowing some records to be traced back to specific individuals; due to unmitigated control gaps, the original de-identification assessment did not account for such correlation capability, requiring reassessment of all existing de-identification work, triggering compliance exposure and operational reputational costs.

Framework Mappings

OWASP Top 10 for LLMLLM02
NIST AI 600-1Data Privacy
MITRE ATLASAML.T0024
ISO/IEC 5338運作與監控
MIT AI Risk RepositoryDomain 2

Risk Treatment & Implementation Guidance

Validate de-identification with adversarial linkage testing simulating re-identification against public information; Control released field combinations under an information-separation principle to reduce linkage surface; Periodically re-evaluate de-identification as externally available information evolves