🤖 AI Summary
This work proposes a novel quantum optical bit commitment protocol to address the fundamental impossibility of unconditionally secure bit commitment in classical settings. By introducing a phase-encoding mechanism and leveraging a physical-layer security assumption—namely, that the network provider ensures communication channels are immune to eavesdropping—the protocol circumvents the well-known Mayers attack under the honest-but-curious model. The authors rigorously establish its security through information-theoretic analysis, thereby providing a provably secure building block for privacy-preserving distributed AI computation. This advancement enables secure multiparty computation with strong guarantees for data confidentiality, marking a significant step toward practical implementations of secure function evaluation in distributed environments.
📝 Abstract
With the rise of artificial intelligence and machine learning, a new wave of private information is being flushed into applications. This development raises privacy concerns, as private datasets can be stolen or abused for non-authorized purposes. Secure function computation aims to solve such problems by allowing a service provider to compute functions of datasets in the possession of a a data provider without reading the data itself. A foundational primitive for such tasks is Bit Commitment (BC), which is known to be impossible to realize without added assumptions. Given the pressing nature of the topic, it is thus important to develop BC systems and prove their security under reasonable assumptions. In this work, we provide a novel quantum optical BC protocol that uses the added assumption that the network provider will secure transmission lines against eavesdropping. Under this added assumption, we prove security of our protocol in the honest but curious setting and discuss the hardness of Mayer's attack in the context of our protocol.