Laconic attribute-based PSI on authenticated inputs and applications

Private Set Intersection (PSI) enables two parties to compute the intersection of their private sets without revealing any additional information. However, existing PSI protocols either fail to guarantee input authenticity-allowing malicious parties to inject arbitrary elements-or lack fine-grained access control, requiring parties to remain online for each query. In this work, we introduce LAB-PSI-AI , a laconic attribute-based PSI protocol that simultaneously achieves authenticated inputs and fine-grained access control. Technically, we employ a BLS signature scheme to guarantee input integrity, a functional commitment for monotone span programs (MSPs) to enforce access policies, and a subvector commitment to achieve laconic communication, where the receiver’s message size is independent of its set. The receiver can recover the payload only if: (i) the item belongs to the database, (ii) a valid signature on the item exists, and (iii) the associated attribute vector satisfies the policy. As an application, we demonstrate a privacy-preserving attribute-based content moderation system. We prove the security of our protocol in the random oracle model under the N -wBDHE and DP-BDHE assumptions.Compared with representative prior PSI schemes, our protocol is the first to simultaneously achieve laconic communication, authenticated inputs, and fine-grained access control, while preserving constant-size messages and constant receiver-side decryption. Experimental results further show that our scheme exhibits constant decryption time in practice and that the sender’s running time remains independent of the number of database items, making it particularly suitable for resource-constrained devices.