Vitalik Buterin said AI agents will replace crypto wallets and front-end interfaces while blockchains coordinate interactions.
Ethereum News
Ethereum (ETH) co-founder Vitalik Buterin argues that artificial intelligence will take over the role of wallets, interfaces, and front-end applications across blockchain networks, while humans remain the underlying principals. "I personally still think of people as being the users," Buterin said in a recent interview. "I think of AI as being the replacement for UI."
Blockchain as a Coordination Layer for AI
In environments where multiple AI systems built by different developers must interact without central oversight, on-chain infrastructure provides the only viable trust and settlement layer, Buterin argued. Incentive structures enforced by smart contracts, including rewards and penalties, are the mechanism that makes decentralized AI coordination possible, he said. "If you have more decentralized AI, that means you have different AI things, agents, programs, whatever you call it, that are built by different people, controlled by different people, that need to interact with each other," Buterin said.
On wallet infrastructure, Buterin predicted that traditional wallet interfaces could eventually disappear entirely. AI systems would instead interact directly with wallet software development kits (SDKs) and application programming interfaces (APIs), removing the need for a visual front-end. "Maybe you don't need a wallet," he said. "You just talk to the SDK." He added that many AI-driven workflows may also tolerate slower execution times compared to current user expectations, particularly when agents are performing complex, multi-step reasoning tasks.
Zero-knowledge (ZK) proofs emerged as another key theme. Buterin said AI agents and users should be able to verify credentials, reputation, and fund sources without fully revealing their identities. "You can make zero-knowledge proofs about your reputation," he said. "You can make zero-knowledge proofs about the source of your funds." The goal, he said, is selective disclosure, where an interaction requires only the minimum information necessary.
Layer-2 Design and ZK Payments
Buterin also warned that AI creates structural vulnerabilities in existing governance mechanisms. Many systems previously protected by the effort required to attack them could be compromised automatically once AI lowers that barrier. "A lot of things that used to require a high amount of effort to break, you can break them automatically," he said.
On layer-2 (L2) design, Buterin said current L2 networks have largely copied the Ethereum Virtual Machine (EVM), scaled it up, and increased centralization before anchoring data back to layer-1 (L1). He argued instead that future L2 systems should be built around specific application requirements, such as privacy, trading, or specialized execution environments. At the protocol level, he identified ZK payments as the likely next step for native AI agent payment standards.
The Ethereum team is building a ZK application programming interface (API) designed to separate each agent request from every other, protecting user privacy without requiring a separate on-chain transaction for each action. The system can be combined with security deposits to deter abuse, allowing AI agents to make payments and requests anonymously across sessions without linking activity to a persistent identity.
