a16z and Dan Boneh: AI's math leap could compress a decade of crypto trust-building into one month

a16z · x · 2026-10-14

a16z crypto's Eddy Lazzarin and Stanford cryptographer Dan Boneh discuss how AI's improving math abilities are disrupting cryptography's trust model.

Cryptographic schemes traditionally rely on time: they're published and then wait years while cryptographers and mathematicians try to break them.
Boneh notes AI changes this: assumptions that used to take 10 years to validate can now be stress-tested by letting a model run for a month; if it can't break the scheme, confidence rises sharply.
The discussion covers fact vs. assumption, risks cryptographers have long prepared for, where quantum computing does and doesn't fit, and opportunities for builders.
Lazzarin warns a whole new well of cognitive ability has appeared to scrutinize these schemes, and the luxury of time is gone.

More from AGI Musings