Vitalik Buterin pushed back against the view that AI hackers will cause cybersecurity to collapse. He said cybersecurity is defensive by nature if it is handled seriously, and that he continues to hold crypto, about 90% of his net worth, as a bet on that view. According to PANews, Buterin said that if AI can prove the Navier-Stokes equations and Fermat's Last Theorem, it should also be able to prove that a program is secure as a mathematical theorem, even when the program is complex.

Buterin said the definition of security is itself complex and includes multiple attack scenarios such as forgery, replay attacks, server compromise, key leakage, and side-channel attacks. He said the definition is still smaller than the attack surface of the code and is easier to verify. He added that past efforts that only verified parts considered security-critical often failed.

Buterin said that with AI, the entire program must be verified, including the database, network, cache, and all other parts. He said this is not about good actors finding vulnerabilities before bad actors, but about writing code that is more resistant to attacks. He said this is the direction of Ethereum's development over the next few years. Buterin added that without this approach, blockchains, especially those that combine scalability and privacy features, will have no future.