- Paper 1: Privacy-preserving attribute verification replaces ID+selfie age checks with unlinkable tokens attesting only the required predicate (e.g., over 18), issued by an M-of-N threshold committee to avoid single points of control, and eliminates vintage leakage through stable predicate keys or mixing.
- Paper 1 also uses on-device liveness gates to prevent replay and deepfake attacks, ensuring biometric data never leaves the device.
- Paper 2: Publishing group voting results (e.g., by demographic) without exposing individuals by using Fréchet bounds to measure per-cell privacy risk, adding random noise only when margins near extremes would reconstruct a voter's choice.
- Paper 3: A remote voting system that simultaneously achieves secret ballot, verifiable integrity (without trusting the operator), and continuous tally, using cryptographic building blocks and closing coercion via fake credentials.
- Paper 3 addresses device malware with a dual-root design (attested execution + biometric-gated device) and provides a deployability inequality: only safe when the value of outcome manipulation is less than the cost of subverting both roots.
- All three papers replace institutional trust (ID verifier, data collector, election authority) with verifiable proofs and provide a computable number that indicates where the substitution is safe.