Cross-Silo De-Anonymization Under Local Differential Privacy: Threat Model, Phase Transition, and Coordination Necessity
arXiv:2606.16763v1 Announce Type: cross Abstract: When a person's records appear in k independent data silos, each protected by (epsilon, delta)-differential privacy, standard composition yields a valid (k*epsilon, k*delta)-DP guarantee for the joint output. This worst-case bound, however, does not answer the concrete inference question: at what k can an adversary actually identify a target person? This paper develops the information-theoretic framework needed to answer that question. We introduce cross-silo person-level DP (XSP-DP), a Pufferfish-style privacy notion whose adjacency relation c
The proliferation of distributed data and the increasing demand for secure data sharing, alongside privacy concerns, necessitate new privacy frameworks like XSP-DP to withstand sophisticated de-anonymization attacks.
This paper presents a formal framework for understanding the privacy risks of de-anonymization across multiple data silos, which is critical for designing robust privacy-preserving systems and policy.
The understanding of differential privacy's limitations in a cross-silo context is enhanced, requiring more sophisticated approaches to protect individual privacy even with seemingly strong local protections.
- · Privacy researchers
- · Data privacy solution providers
- · Individuals with privacy concerns
- · Organizations with siloed, de-identifiable datasets
- · Adversaries attempting de-anonymization
Increased awareness and demand for advanced cross-silo privacy protection mechanisms.
Development of industry standards and regulatory requirements for 'cross-silo person-level DP' in data sharing agreements.
A shift towards privacy-by-design architectures that intrinsically account for multi-silo de-anonymization risks, impacting data collection and sharing practices across sectors.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG