Vitalik Buterin Proposes AI "Stewards" to Automate DAO Voting Across Thousands of Governance Decisions

Vitalik Buterin proposed a framework in which personal AI agents, trained on individual user values, could automate voting across thousands of DAO governance decisions. The design uses zero-knowledge proofs and secure computation to protect voter identities and prevent bribery, while prediction markets would filter spam proposals and AI agents would escalate only critical issues for human review. Buterin said personal LLMs could address the "attention problem" created by high governance volumes facing DAO participants.