Vitalik Buterin dismisses the idea that AI-powered hacks doom cybersecurity
Ethereum co-founder argues that artificial intelligence will make it possible to verify code security through mathematical proofs.
Ethereum co-founder Vitalik Buterin rejected the thesis that artificial-intelligence (AI)-enhanced hacks leave cybersecurity without options. “The idea that hacking with AI means cybersecurity is doomed is becoming increasingly common. I don’t agree.”
“Because cybersecurity naturally favors defense once people get serious,” Buterin said, adding that “anyone who continues to hold cryptocurrencies (including me, ~90% of my net worth) is implicitly making that bet,” the developer said yesterday, September 16, 2026.
Buterin’s central point is that AI can treat software security as a mathematical theorem, even when dealing with complex code. “If the AI can prove Navier-Stokes and FLT, then the AI can prove the statement ‘this program is secure’ as a mathematical theorem. Even if the program is very complicated,” he argued.
The acronym FLT refers to the Last Fermat Theorem, while Navier-Stokes corresponds to a set of differential equations from mathematical physics; both represent complex proofs.
The developer warned that the word “secure” hides “all kinds of skeletons in the closet in terms of what it really means.” To illustrate this complexity, he used the messaging app Signal as a test case. He explained that the most obvious definition—preventing someone without the recipient’s key from reading the messages—does not cover scenarios where the adversary can forge messages, block their delivery, or actively attack the system.
$BTC $ETH $ZEC
Ethereum co-founder argues that artificial intelligence will make it possible to verify code security through mathematical proofs.
Ethereum co-founder Vitalik Buterin rejected the thesis that artificial-intelligence (AI)-enhanced hacks leave cybersecurity without options. “The idea that hacking with AI means cybersecurity is doomed is becoming increasingly common. I don’t agree.”
“Because cybersecurity naturally favors defense once people get serious,” Buterin said, adding that “anyone who continues to hold cryptocurrencies (including me, ~90% of my net worth) is implicitly making that bet,” the developer said yesterday, September 16, 2026.
Buterin’s central point is that AI can treat software security as a mathematical theorem, even when dealing with complex code. “If the AI can prove Navier-Stokes and FLT, then the AI can prove the statement ‘this program is secure’ as a mathematical theorem. Even if the program is very complicated,” he argued.
The acronym FLT refers to the Last Fermat Theorem, while Navier-Stokes corresponds to a set of differential equations from mathematical physics; both represent complex proofs.
The developer warned that the word “secure” hides “all kinds of skeletons in the closet in terms of what it really means.” To illustrate this complexity, he used the messaging app Signal as a test case. He explained that the most obvious definition—preventing someone without the recipient’s key from reading the messages—does not cover scenarios where the adversary can forge messages, block their delivery, or actively attack the system.
$BTC $ETH $ZEC
