A team of researchers from Cornell University, including Ethereum co-founder Vitalik Buterin and PhD students Mahimna Kelkar, Kushal Babel, Philip Daian, and James Austgen, is delving into the potential risks that decentralized autonomous organizations (DAOs) may face as they become more mainstream. Their focus is on the looming threat of unified attacks on protocols through smart contract bribery, which could lead to the emergence of “dark” voting systems.

During the Science of Blockchain Conference at Columbia University in early August, Mahimna Kelkar spoke about the group's groundbreaking research on proofs of complete knowledge (CK), a novel cryptographic concept they introduced in 2023.

Kelkar explained the foundational idea behind proof of knowledge, a widely used cryptographic method in the crypto industry that allows one party (the prover) to convince another party (the verifier) that they possess secret information, such as a secret key, without actually revealing the information. However, the team identified a significant vulnerability in this concept. Kelkar noted:

“When the secret key is held inside trusted hardware, in what we call encumbrance of the secret key, you can still complete this proof of knowledge without actually having knowledge of the underlying secret key.”

This subtle gap in standard proofs of knowledge definitions could leave voting protocols susceptible to bribery attacks, a significant concern in DAOs where governance is decentralized. Kelkar elaborated,

“A voting platform may be vulnerable to bribery attacks, where users can sort of sell their votes to bribers in a dark marketplace. What our work tries to do is establish an individual, real person kind of ownership of data.”

The researchers proposed two methods to enforce proof of complete knowledge, thereby mitigating the risk of these attacks. One approach involves using a trusted execution environment (TEE) to prove that a voter possesses a key and can use it. Even if an attacker attempts to lock the key away, the key remains under the control of the voting system’s own TEE, ensuring that the tokenholder maintains complete control.

The second method involves using application-specific integrated circuits (ASICs), typically employed in Bitcoin mining. By sending a key to an ASIC, which lacks a TEE environment, the key remains accessible to the user, ensuring they retain control while still proving that the key was used by the ASIC.

While this research is still in the prototype stage, the team has demonstrated that these threats are realistic. Kelkar concluded with

"We show that this is a realistic threat to DAOs, and we show this by demonstrating a practically deployable dark DAO, which can facilitate vote buying in existing DAOs. It’s not something that you can deploy tomorrow, but it’s like practically instantiable as a research prototype today.”