Vitalik Buterin sees AI-assisted verification as path to safer Ethereum code

Traders Union
All News
Crypto News
AI-assisted verification to safe Ethereum code

Highlights

Buterin published an article on the use of formal verification in blockchain security.
Ethereum frontier research is developing an approach where code can be written directly in EVM bytecode, assembly or Lean.
Buterin believes AI could improve both development efficiency and code security.

Code that can be proven correct

Buterin described a new paradigm emerging in Ethereum frontier research: developers can write code at a low level, including EVM bytecode, assembly or Lean, and then confirm its correctness through mathematical proofs that are automatically checked in Lean.

The point of this approach is to replace part of the traditional reliance on developers, auditors and tests with strict verification of a program’s properties. If the proof is correctly written and checked, the system receives a stronger guarantee that the code does exactly what it is supposed to do. Researcher Yoichi Hirai calls this model the “final form of software development.”

Where AI can help

Buterin believes AI-assisted formal verification can improve both code efficiency and security. This is especially important for parts of Ethereum where a single error could cost hundreds of millions of dollars or damage trust in the infrastructure. Among such modules, he highlights STARKs, ZK-EVMs, post-quantum signatures and consensus algorithms.

In this model, AI should not simply generate code without oversight. Its role is closer to that of an assistant that speeds up proof writing, helps find errors and checks whether code matches its specification. For Ethereum, this fits into a broader development path: the network is increasingly positioned not as the fastest blockchain platform, but as a base layer where security and verifiability matter more than the speed of individual transactions.

The line between core and periphery

Buterin also stresses that formal verification is not a universal solution. It can still fail if proofs do not cover all cases, if the specification itself is written incorrectly, or if a vulnerability appears through hardware side channels.

That is why the future architecture of software may split into two parts: small “security cores” that go through the strictest possible formal verification, and less critical peripheral components where more flexible development methods remain acceptable. In such a model, Ethereum could become one of the key security cores of the digital economy. If this approach spreads, the main shift will not be that code is written faster, but that the most important code becomes harder to break.

As we previously reported, Vitalik Buterin proposes simplifying Ethereum node architecture.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.