Vitalik Buterin offers human-validated AI wallets for crypto transactions

Buterin proposed AI-assisted wallets where algorithms suggest transaction plans, but users must manually confirm large transfers.

Vitalik Buterin shares his vision of how artificial intelligence (AI) can redefine the next generation of Web3 wallets.

He also proposed a model where people are directly involved in validating high-value transactions.

AI is shaping new crypto wallets

The founder of Ethereum has shared his views on decentralized social media platform Farcaster, noting that it is “absolutely clear” that the next iteration of wallets will include AI.

However, Buterin added that he does not trust LLMs with multimillion-dollar transactions and control over large sums of money. Instead, he proposed an approach in which AI systems assist users while leaving the final decision in human hands.

He described an optimal workflow in high-value situations that involves an AI system proposing a plan and then a local light client simulating the transaction. The person then reviews the intended action and desired outcome before manually approving.

However, Buterin cautioned that this approach should be applied conservatively with a strong focus on security. He suggested that one way to achieve this is to remove decentralized application interfaces from the transaction process. By removing dApp user interfaces from the flow, the system can reduce several attack vectors related to theft and privacy risks.

The 32-year-old has previously discussed how cryptocurrency and AI can evolve together. He envisions blockchains and technology working together, and crypto provides the trust, privacy, and economic infrastructure it needs to operate securely and fairly.

You may also like:

AI-powered proposed wallet workflow

Other developers and community members responded to Buterin’s comments by describing possible implementations of the idea.

Ethereum developer Andrey Petrov proposed two additional scenarios. Initially, the user initiates the transaction normally while the AI ​​analyzes the signed payload. The technology then tries to guess their intended action and interpret it in plain language, allowing them to confirm whether the transaction accurately reflects what they intended.

In the second case, the user either directly states his intended action or relies on the explanations generated in the first step. The AI ​​then tries to reset the transaction on its own, without reference to the original amount, to determine if it achieves the same result. He explained that any differences between the two indicate areas that require further consideration before the process is finalized.

Another Farcaster user, identified as fkaany, described a framework in which AI plans complex crypto strategies such as multi-hop swaps, yield optimization, and gas minimization.

This includes a local light client that simulates output, allowing individuals to review the exact summary and manually confirm the transaction, helping to reduce risks from blind signatures, phishing interfaces, and malicious dApp payloads.