Oct 13, 2026 · 42 min
AI Raises Cryptography Questions Without Breaking Encryption
Can AI Break Cryptography? What the Experts Actually Think
The episode separates a speculative AI risk from the established quantum threat while outlining how institutions can prepare for failures in public-key cryptography.
- 1Recent AI advances have not demonstrated an attack that breaks current encryption or elliptic-curve cryptography.
- 2AI could accelerate cryptographic review, exposing weaknesses sooner while also testing and strengthening security assumptions.
- 3Institutions should maintain current defenses, monitor developments, and plan hybrid migrations and backup algorithms for future failures.
Don't miss
The panel explains how a confirmed elliptic-curve break would trigger disclosure, coordinated migration, and wider adoption of hybrid protections.
The brief
The episode asks whether AI could undermine the cryptography protecting banking, communications, blockchains, and wallets, while distinguishing speculation from the established quantum threat.
Dan Bonet and the panel stress that recent AI breakthroughs have not broken current encryption; they may instead accelerate the testing of cryptographic assumptions.
A confirmed break of elliptic-curve cryptography would reach far beyond blockchains, affecting banking, healthcare, government communications, and much of digital infrastructure.
The practical response is preparation rather than panic: keep using current systems, maintain security practices, and develop hybrid defenses and backup algorithms.
The standout tension is that large-scale AI testing could either uncover a devastating attack or increase confidence by failing to find one.