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

Vitalik Buterin has outlined his perspective on how artificial intelligence (AI) could redefine the next generation of Web3 wallets.

He also proposed a model where humans remain directly involved in approving high-value transactions.

AI Will Shape Newer Crypto Wallets

The Ethereum co-founder shared his views on the decentralized social media platform Farcaster, noting that it is “pretty obvious” that the next iteration of wallets will heavily involve AI.

Despite this, Buterin added that he would not trust LLMs with multi-million-dollar transactions or control over large amounts of money. Instead, he gave 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 would involve an AI system proposing a plan, after which a local light client simulates the transaction. The person would then review the intended action and the required outcome before manually confirming it.

However, Buterin warned that this approach must be implemented conservatively with a strong emphasis on security. He suggested that one way to achieve this is by removing decentralized application interfaces from the transaction process. By eliminating dApp user interfaces from the flow entirely, the system could reduce several attack vectors associated with theft and privacy risks.

The 32-year-old has previously discussed how cryptocurrency and AI could evolve together. He envisions blockchains and the technology working hand-in-hand, with crypto providing the trust, privacy, and economic infrastructure that it needs to operate safely and fairly.

You may also like:

Proposed AI-Assisted Wallet Workflows

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

Ethereum developer Andrey Petrov suggested two additional scenarios. In the first, a user initiates a transaction as usual while AI analyzes the payload about to be signed. The technology would then attempt to guess their intended action and explain it in plain language, allowing them to confirm whether the transaction accurately reflects what they meant to do.

In the second case, the user either states their intended action directly or relies on the explanation generated in the first step. The AI then tries to reconstruct the transaction independently, without referencing the original amount, to determine whether it arrives at the same outcome. He explained that any differences between the two would show areas that require further review 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 would involve a local light client simulating the outcome, which would allow individuals to review a clear summary and manually confirm the transaction, helping reduce risks from blind signing, phishing interfaces, and malicious dApp payloads.

TLDR:

• Minimmit achieves finality in one signing round, replacing Casper FFG’s two-round justification and finalization process. (truncate to fit — 105 chars)
  
• The new gadget lowers fault tolerance from 33% to 17%, but raises the unilateral censorship threshold from 67% to 83%.
  
• Buterin argues censorship poses a greater threat than finality reversion, as it lacks immediate, verifiable on-chain evidence.
  
• Minimmit requires 83% of clients to share a bug before incorrect finalization occurs, giving developers a wider safety margin.

Minimmit has been put forward as a direct replacement for Casper FFG within Ethereum’s consensus layer. Ethereum co-founder Vitalik Buterin recently shared a detailed technical post comparing both finality gadgets.

Casper FFG has long served as a two-round finality mechanism on the network. The proposed system, by contrast, achieves finality in a single round of validator signatures.

The proposal is drawing attention as the Ethereum community continues to evaluate changes to its consensus architecture.

Why the New System Operates in a Single Round

Casper FFG asks each attester to sign a block on two separate occasions. The first signature “justifies” the block, and the second “finalizes” it.

Minimmit cuts this down to a single signing round. This makes the process more efficient for validators across the network.

The change comes with a direct cost to fault tolerance, though. The new system’s threshold sits at 17%, compared to 33% under Casper FFG.

A smaller portion of malicious stake can therefore disrupt finality under the new model. Still, Buterin’s post makes the case that other properties of the system more than offset this drop.

In the post shared on X, Buterin described himself as a long-standing “security assumptions hawk” in Ethereum’s consensus research. He cited his past push for 49% fault tolerance under synchrony.

He also referenced his work on DAS for dishonest-majority-resistant data availability checks. Despite this record, he stated he is “even enthusiastic” about the proposed design.

The asynchronous network case also differs between the two systems. Under ideal 3SF, finality holds as long as an attacker controls less than 33% of stake.

The proposed gadget lowers that same protection to 17%. In both cases, any reversion of finality triggers massive slashing penalties against offending validators.

Censorship Resistance and the Broader Security Picture

Buterin’s argument centers on identifying censorship as the more dangerous threat. Unlike finality reversion, censorship produces no immediate, publicly verifiable evidence against the attacker.

A reversion event, on the other hand, results in automatic, large-scale slashing. This asymmetry is a core reason behind his support for Minimmit’s design.

Both systems require an attacker to control over 50% of staked ETH to carry out censorship. The key distinction lies in what happens at higher thresholds.

In 3SF, an attacker above 67% can finalize the chain unilaterally, removing any coordination point for honest validators. The new system raises that threshold to 83%.

Software bugs present another area where the proposed gadget holds an advantage. Under 3SF, a flaw shared by 67% of client software can accidentally finalize an incorrect chain state.

Minimmit raises that bar to 83%. This wider margin gives developers more time to identify and respond before errors become permanent.

Buterin also addressed the economic argument against finality reversion attacks. With 15 million ETH staked, reverting finality under 3SF would require slashing 5 million ETH, or roughly $10 billion.

He noted that the 17% baseline still represents an enormous deterrent on its own. From there, he argues the proposed system’s other properties make it the stronger overall consensus design for Ethereum.