Ethereum co-founder Vitalik Buterin has proposed a technical overhaul of decentralized autonomous organizations (DAOs), calling for the use of personal artificial intelligence agents that privately vote on behalf of users and help expand digital governance.
The plan was unveiled on social media platform X a month after Buterin criticized the DAO for low participation and centralization of power, and is aimed at steering users away from delegating votes to large token holders.
Instead, individuals will deploy their own AI models, trained on past messages and stated values, to vote on the thousands of decisions facing the DAO.
“Thousands of decisions must be made, involving many areas of expertise, and most people do not have the time or skills to become experts in any one area, let alone all of them,” Buterin wrote. “So what can we do? We use a personal LLM to solve attention problems.”
The first is content privacy, which ensures the confidentiality of sensitive data. AI agents operate within secure environments such as multi-party computing (MPC) and trusted execution environments (TEEs), allowing them to process private data without exposing it to public blockchains.
The second is participant anonymity. Buterin called for the use of zero-knowledge proofs (ZKPs), cryptographic tools that allow users to prove they are eligible to vote without revealing their wallet address or voting method.
This prevents coercion, bribery, and whale watching, where a small number of voters copy the decisions of large numbers of token holders.
These AI stewards automate day-to-day governance participation and flag only important issues for human review.
To weed out low-quality or spammy suggestions, an emerging problem as generative AI floods open forums, Buterin suggests launching a predictive market. In this case, the agent can bet on the possibility that the proposal will be accepted.
Good bets earn dividends and encourage valuable contributions while penalizing noise.
Buterin also called for privacy-preserving tools such as multi-party computing and trusted execution environments that would allow AI agents to evaluate sensitive data, such as job applications or legal disputes, without exposing them to public blockchains.
Read more: From 2016 hack to $150 million donation: DAO’s second law will focus on Ethereum security
