Buterin Disagrees: Improving AI Won't Doom Cybersecurity

Ethereum co-founder Vitalik Buterin recently expressed his disagreement with the notion that improving artificial intelligence (AI) will eventually make cybersecurity unwinnable. In a lengthy social media post on X, Buterin argued that AI could instead give defenders the tools needed to build software significantly harder to exploit.

Buterin tied this belief to his own exposure to cryptocurrency, stating that roughly 90% of his net worth remains in crypto and that anyone holding substantial amounts of it is implicitly betting that secure digital systems can be built despite sophisticated attackers. He also referenced formal verification, a method of mathematically demonstrating software satisfies specific properties.

The Ethereum co-founder used the encrypted messaging app Signal as an example to illustrate why proving software security might be challenging. An attacker could attempt various methods to compromise the system, and even if the content remains encrypted, they might still learn who is communicating. Buterin pointed out that defining what 'secure' means in such cases can become extremely complicated.

Buterin acknowledged that more capable AI systems could make finding vulnerabilities easier for attackers. However, he argued that this improvement in machine reasoning could also lower the cost of formally proving software behaves according to security specifications. Both sides will have the advantage of AI, making cybersecurity a potentially winnable endeavor.