Ethereum

AI Could Crack Crypto Keys Before Quantum, Justin Drake Warns

The Ethereum researcher says the worst case is "months not years". No such break has been shown.

⏱ 2 min read Ethereum
Quick Summary
  • Drake defines a break as recovering a private key in about a week on a large GPU cluster
  • His fix needs no new cryptography: move funds to fresh addresses whose public keys stay hidden
  • In March he put the odds of a quantum computer breaking crypto keys by 2032 at 10% or more

Ethereum Foundation researcher Justin Drake has warned that AI could break the signature scheme securing Bitcoin and Ethereum wallets before quantum computers do.

In a post on X on Wednesday, Drake said it is now reasonable to prepare for ECDSA, the elliptic curve signature scheme both networks use, to fall before Q-Day, the point at which a quantum computer can break today’s cryptography. In the worst case, he said, that could happen “in months not years.”

What he means by a break

Drake defined a break as recovering a private key from a public key in about a week, using hardware that is already available, such as a large GPU cluster. Anyone who could do that could take the funds in any wallet whose public key is visible on-chain.

He pointed to a set of 722 mathematical results that OpenAI published on Tuesday as a sign of how fast AI-driven maths is moving. Elliptic curves, he argued, “feel especially vulnerable to superintelligence,” because their mathematical structure leaves room for clever attacks that hash functions are designed to resist.

What he says to do

Drake’s advice is not to panic. The defence needs no new cryptography or new wallets.

A wallet’s public key is only exposed once it has sent a transaction. Funds held at a fresh address, where the public key is still hidden behind a hash, are much harder to attack. Drake called this “bunker mode” and suggested moving funds to such addresses in a calm, controlled migration rather than a rush.

He also called on large custodians, exchanges and Tether to harden their cold storage, and said he would push for Ethereum to accelerate its move to hash-based cryptography.

How seriously to take it

This is one researcher’s worst-case view, not a demonstrated attack. Nobody has shown AI recovering an ECDSA private key, and no other expert has publicly backed the “months” timeline.

It does fit a pattern in Drake’s thinking. In March, after a Google paper on quantum threats, he put the odds of Q-Day arriving by 2032 at 10% or more.

Bitcoin has one unusual buffer. About 20,000 early addresses linked to Satoshi Nakamoto each hold 50 BTC with exposed public keys. Drake expects any attacker to target those first, which would give everyone else warning.

⚖️ Our Verdict ⚖️ Watch and Wait

This is one researcher's worst-case warning, not a demonstrated attack, and no other expert has backed the timeline. The precaution is cheap and needs no new tools, which makes it worth knowing about without being a reason to panic.