Deployments Cursive Consumer Private Overlap
Post
Cancel

Cursive Consumer Private Overlap

Cursive built demos of private set intersection between two consumer’s sets of personal data: contacts, event attendance, emails, personality quizzes, hiring compatibility and more.

The initial demos used a combination of MPC and FHE implemented in https://github.com/RiverRuby/2P-PSI. It uses interactive multi-party BFV to do a dot-product between two bitvectors to discover overlap. This was deployed at ZKSummit11 over contacts data collected with NFC, and the Signature Singularity Residency over event attendance data collected from your email.

The next demo used another combination of MPC and FHE implemented in https://github.com/cursive-team/pz-hiring. It uses non-interactive multi-party FHEW using the Phantom-zone VM, which enabled more complex overlap logic. Specifically it privately checks if a candidate’s salary requirements are below a recruiter’s salary budget, detecting compatibility without revealing other information.

The final demo used a simple scheme involving computing a HMAC of the data label alongside a shared secret generated with ECDH. This serves as a PRF for the data. The HMACs for a pair Alice and Bob would be stored on a server, and a server would be trusted to check which HMACs overlap without colluding with either party to break the other party’s privacy. This was implemented in https://github.com/cursive-team/connections.

Code

Presentation

This post is licensed under CC BY 4.0 by the author.

In recent years, MPC (secure multi-party computation) has seen an increase in adoption in real-world use cases. Find deployments here and add more.


Recently Updated
Contents