Detecting Rogue Decryption in (Threshold) Encryption via Self-Incriminating Proofs
Event details
| Date | 27.05.2025 |
| Hour | 15:00 › 16:00 |
| Location | |
| Category | Conferences - Seminars |
| Event Language | English |
By James Chiang
Abstract
Secure Multi-Party Computation (MPC) is increasingly deployed in high-stakes settings, such as institutional blockchain custody and privacy-preserving machine learning. While security of MPC assumes a threshold of honest parties, rational adversaries will choose to collude when collusion is undetectable. In particular, colluding parties can use an “outer” MPC protocol to “simulate” unauthorized secret share reconstruction or threshold decryption and prevent any trace or proof of collusion from leaving the outer MPC. This talk presents a line of research aimed at mitigating such undetectable collusion. We review the notion of Individual Cryptography [CRYPTO’23] and introduce our follow-up work on Accountable Threshold Decryption, which enables (1) public self-incrimination during decryption and (2) self-incriminating threshold decryption schemes that ensure that colluding parties obtain evidence of collusion, enabling mechanism design that incentivizes the reporting of such behavior.
Prior to academia, James designed flight hardware for Mars rovers at JPL NASA, advised technology firms in three continents at the Boston Consulting Group, co-founded the leading solar business software provider (Eturnity) and contributed to multiple open-source implementations of the original Bitcoin protocol (Core/Libbitcoin).
More information
Abstract
Secure Multi-Party Computation (MPC) is increasingly deployed in high-stakes settings, such as institutional blockchain custody and privacy-preserving machine learning. While security of MPC assumes a threshold of honest parties, rational adversaries will choose to collude when collusion is undetectable. In particular, colluding parties can use an “outer” MPC protocol to “simulate” unauthorized secret share reconstruction or threshold decryption and prevent any trace or proof of collusion from leaving the outer MPC. This talk presents a line of research aimed at mitigating such undetectable collusion. We review the notion of Individual Cryptography [CRYPTO’23] and introduce our follow-up work on Accountable Threshold Decryption, which enables (1) public self-incrimination during decryption and (2) self-incriminating threshold decryption schemes that ensure that colluding parties obtain evidence of collusion, enabling mechanism design that incentivizes the reporting of such behavior.
Bio
James is a cryptography postdoc at Aarhus University, advised by Ivan Damgård. He holds a PhD from the Technical University of Denmark, where he was awarded the DTU Compute Fellowship, and holds a BSc with distinction from UCLA. His research focuses on privacy-preserving techniques motivated by recent blockchain applications and includes practical advancements in post-quantum threshold ring signatures, novel notions of differential privacy for secure multi-party computation (MPC), and the first formal model of miner-extractable-value (MEV). His research has been published in ACM CCS, Financial Cryptography, AFT and recognized with the Sui Academic Research Award.
Prior to academia, James designed flight hardware for Mars rovers at JPL NASA, advised technology firms in three continents at the Boston Consulting Group, co-founded the leading solar business software provider (Eturnity) and contributed to multiple open-source implementations of the original Bitcoin protocol (Core/Libbitcoin).
More information
Practical information
- General public
- Free
Contact
- Host: Prof. Serge Vaudenay