Against those who believe AI could turn cybersecurity into a losing battle, Vitalik Buterin argues the exact opposite. The Ethereum co-founder believes the same technology capable of finding vulnerabilities can be used to mathematically prove that a program is secure. And he says that, by keeping close to 90% of his wealth in cryptocurrencies, he himself is making that bet.
AI's growing ability to find and exploit vulnerabilities has fueled the idea that attackers will end up with the upper hand. Buterin, however, frames the problem the other way around. If AI can prove Navier-Stokes and FLT, then AI can prove the claim 'this program is secure' as a mathematical theorem, he argues.
The key lies in formal verification, a technique that makes it possible to mathematically prove that a program meets certain properties. AI could turn this historically complex and costly procedure into a far more accessible tool for developing software. He argues that there is a prior problem, which is defining exactly what it means for a program to be secure. Buterin uses Signal to explain the difficulty. Encrypting a message to prevent someone without the key from reading it only covers part of the problem.
An attacker could also try to forge messages, block their delivery, compromise the server, tamper with public keys or exploit vulnerabilities in the operating system, the libraries, the compiler or the device itself. Even metadata — who talks to whom, when and how often — can reveal information without any need to break the encryption.
That is why, for Buterin, the great challenge is no longer solely about reviewing millions of lines of code in search of bugs. First you have to describe with enormous precision which properties the system must meet, and then mathematically prove that it meets them.
How AI strengthens cybersecurity in blockchain networks
The proposal also changes the traditional logic of cybersecurity. It is no longer about trusting that the good guys will find every vulnerability before the attackers do. Buterin considers a strategy that makes it possible to build resilient software from the outset and verify its properties to be more robust.
That requires abandoning another common practice: checking only the parts of the program considered critical. According to Vitalik, you have to verify literally the entire program. That includes databases, networks, caches and any other layer capable of introducing a vulnerability. His argument is that modern AI is beginning to make viable something that was previously too costly and difficult.
Ethereum will follow this path
Buterin says Ethereum will move in this direction over the coming years. The reason is especially important for blockchain. Systems that manage economic assets and at the same time want to incorporate greater scalability, privacy and advanced cryptography increase their complexity and also the consequences of any failure.
There is no future for blockchains, especially with scalability and privacy, without doing this, he says. Buterin argues that AI can make hackers more dangerous, but it can also make it possible to build programs with far fewer vulnerabilities to exploit.