Ethereum co-founder Vitalik Buterin has pushed back against claims that artificial intelligence could hand attackers a permanent advantage in cybersecurity. Speaking in a Sept. 17 discussion shared on X, Buterin argued that increasingly capable AI systems could just as easily strengthen defensive software engineering, with his preferred approach centering on formal verification — a method that uses mathematical proofs to test whether software satisfies clearly defined security properties.

The remarks speak to a question now running across the software industry: as AI lowers the cost of probing code for weaknesses, defenders are looking for methods that scale at a comparable pace. The debate carries particular weight for public blockchains, where client code is open and both sides audit the same software at once.

Formal Verification as the Counterweight to AI Attacks

Buterin said AI-driven attacks do not automatically give hackers a lasting edge. His argument rests on formal verification, which applies mathematical proofs to check whether software follows the rules defined for it. As AI systems grow more capable, he said, they could help developers verify increasingly complex programs and test security properties at a larger scale.

The technique's appeal is exhaustive coverage. Conventional testing samples a program's behavior with selected inputs; a completed formal proof covers every execution the underlying model allows, ruling out entire classes of violations at once. That trade — heavier upfront effort in exchange for machine-checked certainty — is what makes verification attractive as AI expands both the volume of code to review and the speed of adversarial probing.

The technique does not, however, guarantee complete security on its own. Developers must first spell out the properties they want the software to satisfy. A proof can confirm that those rules hold, but it cannot cover risks the specification leaves out.

Buterin added that developers need broader checks spanning protocols, servers, databases, networking layers, caches and other supporting components. A formal proof only answers the question developers ask it to test, he noted, and does not identify every missing assumption.

Ethereum Tests AI Agents on Protocol Code

The Ethereum Foundation is already using AI agents in protocol security work. Its Protocol Security team said in July that coordinated agents had found real defects in systems Ethereum depends on. One case involved a remotely triggered crash in the Gossipsub networking code of Rust libp2p — the peer-to-peer networking library Ethereum consensus clients use to exchange blocks and attestations — which developers fixed in version 0.49.4.

The team has also identified a practical weakness in AI-generated security reports. Agents can produce convincing findings that do not affect production systems: some point to unreachable code or testing-only behavior, while others deliver formal proofs that check a weaker condition than researchers intended. The Foundation therefore requires reproduction, automated checks and human review before it accepts a vulnerability.

As AI-generated audit reports multiply, that reproduction-and-review bar serves as the practical dividing line between machine-speed findings and defects worth fixing.

Formal Verification Moves Deeper Into Ethereum Research

Formal verification is set to play a wider role under Ethereum's current protocol research plan. The Ethereum Foundation said on Sept. 7 that the method will support several research tracks through 2029, including privacy, state design, zkEVM development and post-quantum security. The plan also connects verification work with future protocol engineering.

The Foundation expects work on an L1 zkEVM — a design that would verify Ethereum's own block execution with zero-knowledge proofs — to improve verification tools and verified cryptographic components. A separate project, better.codes, combines AI agents with machine-checked proofs in Lean: researchers direct AI systems toward formalized cryptographic problems, while the Lean kernel checks whether each submitted proof meets the stated theorem. The approach pairs faster machine-generated work with strict mathematical checks rather than relying on model output alone.

AI Threats Grow as Ethereum Continues Defense Work

Buterin's comments land as security researchers document the wider use of AI for offensive purposes. Anthropic said its September threat report covered operations in which malicious actors used Claude across cyber activity and other abuse categories from December 2025 through August 2026. The company said some operators used AI to support vulnerability research, exploitation and broader attack workflows.

Security teams also use the same technology to search code, generate tests and reproduce faults faster. Ethereum's own trials show that researchers still need to confirm reachability, production relevance and the exact property that a proof checks.

Ethereum Price Rallies Toward $2,460 Amid the Security Debate

At the time of writing, Ethereum rallied 2.3% over the past 24 hours to $2,457.19. ETH recovered from an intraday dip below $2,400 and moved back toward the $2,460 area. The four-hour chart shows Ethereum trading inside a broader range, with the lower boundary near $2,375 and the upper boundary around $2,570.

Analyst Ali Martinez also noted a possible path toward $3,000 if Ethereum extends the rebound. In a post on X, he said ETH could first move toward the middle of the four-hour range before testing $2,570. According to Martinez, a strong Ethereum price close above $2,570 on the four-hour chart, backed by higher volume, could support a move toward $2,700, followed by $3,000.

This article is for informational purposes only and does not constitute financial or investment advice. Cryptocurrency investments involve risk, and technical targets or security research developments do not guarantee future price performance.