Vitalik Buterin Proposes AI Agents to Automate Ethereum DAO Voting

TLDR

• Vitalik Buterin proposed using personal AI agents to vote on behalf of users in DAO governance
• The system uses zero-knowledge proofs to keep voter identity private and prevent bribery
• AI agents would handle routine votes automatically and only flag important decisions for human review
• Prediction markets would reward quality proposals and filter out AI-generated spam
• Multi-party computation would allow sensitive decisions to be made without exposing private data

Ethereum co-founder Vitalik Buterin has laid out a plan to fix one of crypto’s longest-standing problems: low participation in decentralized autonomous organization (DAO) governance.

Buterin published his proposal on X, about a month after he publicly criticized DAOs for falling into patterns of low voter turnout and power concentration among large token holders.

The core idea is straightforward. Users would deploy their own personal AI model, trained on their past messages and stated values, to vote on their behalf across thousands of DAO decisions.

Buterin argues that the current system is broken by a lack of human attention. Most people don’t have the time or expertise to evaluate every proposal a DAO puts forward.

The typical fix — delegating votes to someone else — makes things worse. Buterin says it creates a small group that controls decisions while everyone else loses influence after clicking the delegate button.

His proposed AI agents would handle routine votes automatically. When the agent is unsure about something important, it would pause and ask the user directly.

Privacy at the Core

A key part of the proposal is keeping voter identity private. Buterin calls for zero-knowledge proofs, a cryptographic tool that lets users prove they are eligible to vote without revealing their wallet address or how they voted.

This is designed to stop coercion, bribery, and “whale watching,” where smaller voters copy the choices of large token holders.

AI agents would run inside secure environments called multi-party computation (MPC) setups or trusted execution environments (TEEs). These allow agents to process private data without it being exposed on a public blockchain.

Filtering Out Bad Proposals

Buterin also addressed the growing problem of spam proposals, which is becoming worse as generative AI makes it easier to flood governance forums with low-quality content.

His solution is prediction markets. Anyone could submit a proposal, and AI agents would bet on whether it would be accepted.

Successful bets would pay out, creating a financial incentive to submit good proposals and penalizing noise.

He also outlined a “public conversation agent” system that would gather views from many participants, summarize them, and identify common ground before allowing responses.

Buterin said good decisions can’t come from simply averaging people’s views. Information must be aggregated collectively first, then participants can respond with full context.

The proposal covers four main tools: personal governance agents, public conversation agents, suggestion markets, and privacy-preserving computation for sensitive decisions like internal disputes or compensation.

Buterin’s post comes as DAOs continue to struggle with getting enough voters engaged to make governance legitimate and functional.

The post Vitalik Buterin Proposes AI Agents to Automate Ethereum DAO Voting appeared first on CoinCentral.