Dan Boneh argues AI can strengthen cryptography by testing it

Dan Boneh: Dan Boneh is a professor at Stanford University who heads the Applied Cryptography Group and co-directs the Computer Security Lab; he also serves as a special research advisor at a16z crypto. In recent discussions hosted by a16z crypto, he addressed concerns about AI potentially breaking cryptography, clarifying that recent AI advances in mathematics have not impacted core cryptographic assumptions. Boneh highlighted how AI can accelerate the process of validating cryptographic schemes by attempting to break them, thereby building greater confidence in their security more rapidly than traditional methods. Math Progress Limits: Recent AI advancements in mathematics and computer science have not produced results related to cryptography problems or undermined existing assumptions. Cryptographic Assumptions: Cryptography security often depends on mathematical problems that are believed to lack efficient solutions, with confidence built through survival against attacks rather than formal proofs. AI Cryptography Validation: Experts note that AI can accelerate validation of cryptographic assumptions by attempting to break them, compressing what traditionally took years into shorter periods.