Justin Drake tells large crypto holders to move to fresh addresses before AI can break ECDSA

Drake's case for a short clock rests on one exhibit. He cited OpenAI's recent mathematical work, including a disproof of the Erdos unit distance conjecture ^[4]. From it he said AI-powered reasoning could break ECDSA "months not years" before qday, the industry's name for the arrival of a quantum computer strong enough to do it ^[15][3]. The exhibit is a result on a different problem. Drake framed the outcome as a potential crisis, and no one has announced that ECDSA is broken ^[16].

On the day, the market priced the warning at roughly zero, judging by the lack of any immediate reaction reported by Crypto Briefing ^[12]. Whether that is the right value depends on the odds of a break within months, and the account of Drake's post does not include an estimate of them.

The cost of acting is uneven. A holder whose coins already sit in an address that has never signed pays nothing extra, because the public key an attacker would need is not visible ^[5]. Exposure concentrates in reused addresses and long-active wallets ^[13], and in custodians, whom Drake wants rotating keys frequently and moving to hash-based multi-signing such as SPHINCS ^[7]. A migration is one transaction. Frequent rotation is a standing line in an operations budget, paid every period whether or not the attack comes.

If no AI result against ECDSA arrives, early movers have paid early for work the quantum transition needs anyway. Hash-based cryptography is already part of Ethereum's long-term roadmap discussions, and Drake has argued for years for moving off traditional cryptographic assumptions ^[11]. If a break lands inside his window, the losses fall on coins whose public keys are already on-chain ^[13]. In the slower version, AI progress shortens the timetable for classical and quantum attacks alike, both of which Drake said recent AI gains should push the industry to discuss faster ^[14], and there is no single day on which anything breaks.

I think the market's zero is defensible on the evidence Drake cited, and the individual move is cheap enough to make anyway. The open question is whether custodians will take on a recurring cost on the strength of one demonstration. The counter-case is that the first public sign of an AI break could be a theft instead of a paper. By then the calm, ordered migration Drake asked for, largest holders first ^[6], is no longer on offer.

A published AI-assisted method that cuts the work of deriving a private key from a public key, the assumption ECDSA rests on ^[3], would show the market wrong. Binance, Bitfinex or Tether starting the migration Drake asked them to lead ^[8] would show it changing its mind.