Key Takeaways:
- Vitalik Buterin claims that AI does not spell the end of cybersecurity, stating that more effective verification may benefit defenders.
- In short, roughly 90% of his net worth is in crypto; security is a personal bet, says Buterin.
- AI aid in formal verification is essential for the security of Ethereum, privacy technologies, and cryptography, he believes.
The worries surrounding the ability of AI to match hacking level have risen, and Vitalik Buterin, one of the founders of Ethereum has addressed the concern. He argued in a new comment that if better math checks were used then critical software could be much harder to exploit.
It’s an increasingly common take that AI hacking means cybersecurity is doomed.
I disagree. I think cybersecurity is naturally defense-favoring once people get their shit together. And anyone who continues to hold cryptocurrency (including me, ~90% of my net worth) is implicitly…
— vitalik.eth (@VitalikButerin) September 16, 2026
Buterin Makes a Security Bet With Crypto
This is basically a bet on the security of digital systems and a wager stronger than ever against those who’ll stop at nothing to compromise it, Buterin explained. He also mentioned that about 90% of his personal wealth is in his crypto portfolio.
He’s not relying on defenders easily locating vulnerabilities sooner than hackers. Rather, he views a more gradual evolution into demonstrating that software abides by defined security rules.
The concept is particularly applicable to blockchains, where a software error can impact wallets, smart contracts, bridges, validators or full networks.
Buterin cited the work of an advanced mathematical theorem proving as what AI may yet be able to do in the future. AI tools that can be very difficult to solve mathematical problems can also be used to solve a formal security specification for a software system, he says.
Read More: Robinhood’s AI Crypto Trading Push Arrives as Robinhood Chain Tops $1B DEX Volume
Formal Verification Could Change Crypto Security
The difference, between proving defined properties and testing code, is that the latter is binary and the former isn’t. Conventional Security Testing involves audit, simulation and attack searches for bugs. Formal verification begins by stating the exact requirement that the system needs to guarantee and then performing a mathematical check to see if the system does.
Buterin pointed out that it wasn’t easy to tackle the security requirement itself. A protocol might be able to prevent unauthorized access, but not stop attacking via forgery, replay attacks, denial-of-service, or failure to a dependent aspect.
Ethereum Needs Whole-System Verification
Verifying just a few so-called critical parts might not be sufficient for crypto networks. Other supporting software, such as databases, networking layers, caches, libraries, and others, can also be included in the attack surface.
Buterin posits that the modern AI might enable quicker validation of a larger part of a full program. That could lessen the need to believe that security researchers will find all the exploitable vulnerabilities before the bad guys.
AI Hackers Are Getting More Capable
There have also been instances where Anthropic has seen the severity of AI systems gaining access to real-world thirdparty systems in the presence of security assessments. That’s what makes automated attacks and automated defences a race.
Buterin’s solution for Ethereum is not just to increase the number of AI-powered scanners. It’s writing software with precise security properties and that are more and more provable by machine.
It could shift the balance in crypto from being more one-sided, where attackers have the upper hand by relying on AI, and more toward the defenders having the edge.
Read More: Anchorage Unveils Agentic Banking, Targeting $1T AI-Driven Crypto Finance Boom
