International Association
for Cryptologic Research

Encrypted search schemes have been proposed to address growing privacy concerns. However, several leakage-abuse attacks have highlighted the shortcomings of these schemes. The literature remains vague about the consequences of these attacks for real-world applications: are these attacks dangerous in practice? Is it safe to use these schemes? Do we even need countermeasures?

This paper introduces a novel mathematical model for attackers' knowledge using statistical estimators. Our model reveals that any attacker's knowledge is inherently noisy, which limits attack effectiveness. This inherent noise can be considered a security guarantee, a natural attack mitigation. Capitalizing on this insight, we develop a risk assessment protocol to guide real-world deployments. Our findings demonstrate that limiting the index size is an efficient leverage to bound attack accuracy. Finally, we employ similar statistical methods to enhance attack analysis methodology. Hence, our work offers a fresh perspective on SSE attacks and provides practitioners and researchers with novel methodological tools.