Justin Drake’s ‘Bunker Mode’ Warning: Why AI Risk to Wallet Cryptography Calls for Calm Preparation
Crypto has spent years arguing about when a quantum computer might be able to break the elliptic-curve signatures that protect Bitcoin and Ethereum wallets. This week, Ethereum Foundation researcher Justin Drake put a different threat on the table: that artificial intelligence, not quantum hardware, could be the thing that cracks those signatures first, and possibly sooner than most people assume.
His proposed response, which he calls “bunker mode,” is not a new wallet or a new cryptographic standard. It is a calm, gradual migration of funds to addresses that have never signed a transaction. The idea has split the community, and it landed on a day when Bitcoin was already sliding toward $82,000. This piece looks at what Drake actually argued, why his recommendation is more modest than the headlines suggest, and where the reasonable middle ground sits.
What Drake Argued
According to reporting from CoinDesk, Decrypt and Cointelegraph, Drake said ECDSA, the signature scheme used by Bitcoin and Ethereum, could be broken before so-called Q-Day, and that in a worst case the timeline could be measured in months rather than years. His evidence was a batch of 722 mathematical results published by OpenAI earlier this week, which he presented as a sign that AI-assisted mathematics is advancing quickly enough to challenge assumptions about elliptic-curve security.
Drake’s specific concern is that elliptic curves have a rich mathematical structure that could make them more exposed to a sufficiently capable AI than hash functions, which are designed to look structureless. He also urged Ethereum to accelerate its planned shift toward hash-based cryptography. The Ethereum Foundation already set up a dedicated post-quantum team earlier this year, so the direction of travel is not new; the urgency is.
Why “Unused Addresses” Matter
The technical core of bunker mode is simple. When you receive funds to a fresh address, the network only sees a hash of your public key. Your actual public key is revealed on-chain only when you sign a transaction spending from that address. If someone ever found a way to derive a private key from a public key, every address that has already signed a transaction and still holds funds would be the exposed set. Addresses that have never signed would still be protected by the hash.
That is why Drake’s advice is to move funds to never-used addresses, and why he stressed that it needs no new wallets or cryptography. Decrypt also noted a point that has circulated in Bitcoin circles for years: roughly 20,000 early addresses linked to Satoshi Nakamoto, each holding 50 BTC with exposed public keys, would likely be the first targets of any attacker, which offers smaller holders a form of indirect cover.
The Pushback
Reaction was mixed. Vitalik Buterin agreed the risk from AI-accelerated mathematics should be taken seriously, but said he did not recommend that anyone scramble to move funds today. Jan3 CEO Samson Mow was blunter, dismissing the warning and arguing there was no reason to panic.
Skeptics have a fair point. Publishing a large set of mathematical results is not the same as demonstrating any weakness in the specific discrete-logarithm problem that secures secp256k1. No one has shown such an attack. “Months not years” is framed by Drake himself as a worst case, not a forecast.
Our Take: Prepare Like It Is a Fire Drill
The useful way to read this episode is not as a prediction but as a reminder that cryptographic assumptions are not permanent. Bunker mode, stripped of the dramatic name, is mostly good hygiene that security-minded holders already practice: avoid address reuse, keep long-term savings in addresses that have not signed transactions, and know how to move funds calmly if a credible break is ever demonstrated.
The real risk of a warning like this is the opposite of complacency. A rushed, panicked mass migration is exactly the environment in which people paste addresses into phishing sites, approve malicious transactions or send funds to the wrong chain. As we noted in our look at Q3 crypto hack losses, most money lost in this industry is lost to operational mistakes and compromised infrastructure, not to broken mathematics.
For long-term holders, the reasonable steps are modest: confirm your hardware wallet generates a fresh receive address each time, understand which of your addresses have already signed transactions, and keep your recovery phrase backed up properly. If you want a refresher on choosing a device, our Ledger vs. Trezor guide covers the basics.
For protocol developers, the message is more pointed. Bitcoin and Ethereum both have post-quantum migration discussions underway, and whether the eventual threat comes from quantum machines or AI-assisted cryptanalysis, the migration path is largely the same. Drake’s warning, overstated or not, is an argument for shortening that timeline rather than for anyone moving coins in a hurry tonight.
This article is for informational purposes only and does not constitute financial advice.
