Skip to content

Can AI Break Cryptography? What the Experts Actually Think

a16z Podcast

Hosted by Erik Torenberg

42m
Transcript

Original podcast

Where to listen

Summary

Written by a model from the transcript; not yet reviewed.

This episode addresses concerns sparked by recent OpenAI math papers and social media posts warning that advanced AI could break modern cryptography faster than quantum computers. The participants, including Stanford professor Dan Boneh and researchers from A16Z Crypto, discuss whether Frontier Labs should test their models against cryptographic assumptions, the implications for blockchain and the wider digital world, and how to balance security risks without panicking. The guest and hosts argue that while AI could potentially accelerate attacks on mathematical assumptions, no such breakthroughs in cryptography have been publicly demonstrated, and existing protocols like post-quantum hybrids and crypto agility provide robust defenses. They explain why everyday users should change nothing, why institutional practices like cold storage are already standard, and how AI can ultimately be used defensively to scan code and eliminate software bugs.

  • Everyday users of technology do not need to change anything or panic because the world of cryptography remains unchanged until new information or concrete attacks are demonstrated. 15:03
  • Institutions should rely on established best practices like crypto agility to quickly switch from one system to another if an assumption is proven insecure. 23:35
  • Frontier Labs should test their models against basic cryptographic assumptions and publicly share the results to provide the tech community with confidence or early warnings. 4:19
  • Startups and software shops can use advanced AI models to continuously scan their codebase and cryptography for bugs, effectively improving software quality. 35:56

Chapters

  1. 0:00
    Podcast Intro and Threat Overview

    Hosts introduce the emerging AI threat to cryptography and set the context.

  2. 0:45
    OpenAI Breakthroughs and Cryptography Concerns

    Exploring how OpenAI’s recent papers sparked fears of crypto breakage.

  3. 2:11
    AI vs Quantum Threats to Cryptography

    Clarifying the differences between AI and quantum computing risks.

  4. 14:36
    AI’s Role in Accelerating Cryptographic Testing

    Using AI models to rapidly evaluate and strengthen cryptographic schemes.

  5. 25:43
    Hash vs Lattice Cryptography Debate

    Comparing hash‑based and lattice‑based systems for security and agility.

  6. 41:05
    Closing Remarks and Listener Guidance

    Hosts wrap up, advise listeners, and provide episode resources.

Highlights

Quotes

  • “The fact that people try to break them and fail gives us more confidence.”
    Dan Boneh · 0:16
  • “The reason we can't prove that those problems are hard is because that's related to a famous open problem in computer science called the P versus NP problem.”
    Dan Boneh · 6:43
  • “Crypto agility is, in some sense, our friend for defending against the unexpected.”
    Dan Boneh · 23:35
  • “The labs say that they failed to break them. That will give us tremendous confidence that our assumptions are valid.”
    Dan Boneh · 8:40
  • A16Z 41:10–41:58

About this episode

Could AI break the cryptography protecting everything from crypto wallets and banking to private messages? Recent advances in AI-driven mathematical research have sparked new warnings about the security of today's encryption systems, but how much of the concern is supported by evidence? a16z crypto's Liz Harkavy is joined by Stanford cryptographer Dan Boneh, a16z crypto General Partner Eddy Lazzarin, and Research Partner Justin Thaler to separate what we know from what remains speculation. They explain why modern cryptography relies on mathematical assumptions, what OpenAI's recent results do and don't demonstrate, and how the potential threat from AI differs from quantum computing. They also discuss why existing security practices and cryptographic alternatives offer reasons for confidence, the importance of building systems that can adapt if an assumption fails, and how AI could ultimately help researchers and developers make software more secure.

Transcript

0:00
42:11

7,493 words · 5 speakers

Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.

Topics

  • cryptography
  • AI
  • lattice
  • hash‑based signatures
  • OpenAI
  • Ethereum
  • Bitcoin
  • Stanford
  • NIST
  • Ethereum Foundation

Page updated · transcript and summary are machine-generated