Can AI Break Cryptography? What the Experts Actually Think

The emergence of advanced AI capabilities has sparked concerns regarding the security of modern cryptographic systems, specifically whether AI models can solve the mathematical problems underpinning digital infrastructure. While recent research from OpenAI demonstrates significant progress in mathematics, no evidence suggests that current cryptographic standards are vulnerable to these models. Cryptography relies on long-standing complexity assumptions, and the industry already maintains "crypto agility" to transition to alternative systems if necessary. Rather than posing an immediate threat, AI serves as a powerful tool for developers to scan code for vulnerabilities and stress-test security assumptions. The current industry focus should shift from "bunker mode" panic to demanding transparency from AI labs regarding their testing results, as these models present a unique opportunity to verify and strengthen the security of global digital systems.