Ethereum Foundation researcher Justin Drake has urged the crypto industry to start preparing for what he calls “bunker mode,” arguing that rapid advances in AI-driven mathematics could threaten cryptographic systems sooner than the industry expects.
In an October 7 post, Drake recommended a controlled migration of assets to fresh addresses whose public keys remain hidden behind hashes. He specifically urged large holders to consider moving funds to addresses that have never signed transactions and to move remaining funds whenever they sign a transaction.
He warned that the industry should consider the possibility that ECDSA could become breakable within “months, not years,” potentially through classical computing rather than quantum computers.
His argument rests on recent advances in AI for mathematics. Drake pointed to breakthroughs involving problems such as integer multiplication, 3SUM, and the Erdős unit distance conjecture as evidence that AI could uncover mathematical shortcuts that humans have missed.
He argued that elliptic curves deserve particular scrutiny because of their mathematical structure and suggested that the industry should accelerate its move toward hash-based cryptography.
Vitalik Takes the Warning Seriously
Ethereum cofounder Vitalik Buterin pushed back against the idea that users should immediately scramble to move their funds, but he agreed that AI-driven advances in mathematics deserve serious attention.
Buterin argued that the industry should consider AI vulnerability alongside quantum vulnerability. He also questioned the assumption that lattice-based cryptography will remain safe simply because elliptic curves face greater scrutiny today.
The programmer still warned users against rushed migrations, warning that he had “lost more money in botched migrations than I have lost in all hacks combined.” He later clarified that multisig users should ideally change signing keys after each operation while continuing to gather signatures offchain where possible.
Not Everyone Sees an Emergency
The response from cryptographers has been sharply divided. Yehuda Lindell, head of cryptography at Coinbase, called Drake's argument a bad take, saying that there is currently no evidence that decades-old assumptions behind elliptic curve cryptography have failed.
Lindell argued that AI proving difficult mathematical theorems does not show it can solve problems like elliptic curve discrete logarithms. He also noted that a genuine break of elliptic curve cryptography would create consequences far beyond crypto, affecting bank websites, financial systems, software signing and other internet infrastructure.
Longtime Ethereum developer and researcher Dankrad Feist made an even more direct point. If attackers can immediately compromise any exposed public key, he argued, moving funds into protected addresses wouldn't necessarily save them because the underlying assets could lose all value.
Dragonfly managing partner Haseeb Qureshi took the opposite view, describing Drake's warning as a sober response to rapid mathematical progress and suggesting the industry should temporarily adopt a more defensive approach.
How Much Cryptography Could AI Already Be Hiding?
The discussion has also shifted toward a harder question: how many novel attacks on cryptography could already exist but remain undisclosed?
Responding to a question about how many such attacks might emerge, Matthew Green, an associate professor of computer science at Johns Hopkins University, raised an even broader concern, saying he thinks the industry could “lose public key cryptography,” particularly public-key encryption.
Solana cofounder Anatoly Yakovenko pushed back on that premise, arguing that AI will not simply solve discrete logarithms, prime factorization, or the P ≠ NP conjecture, but could instead uncover vulnerabilities in specific curve parameters or groups.
That distinction sits at the heart of the debate. AI could potentially discover weaknesses in how cryptographic systems are implemented without overturning their underlying mathematical assumptions.
Solana Says Its Design Offers “A Defined Path Forward”
Solana Foundation VP of Technology Jacob Creech rejected the need for Justin Drake’s proposed “bunker mode” on Solana.
Creech argued that Solana's use of Ed25519 gives users a different recovery path if an attacker eventually breaks the underlying elliptic curve. The signing key derives from a secret seed through hashing, while the seed itself never appears onchain.
According to Creech, a network upgrade could let legitimate owners prove knowledge of their seed with a hash-based proof and migrate their accounts to a safer signature scheme. In the unlikely event of a break in Solana’s cryptography, validators, not users, would be responsible for upgrading infrastructure to ensure continuity. Creech also stated that “a working proof of concept already exists”, pointing to Anza’s Proof of Knowledge Seed protocol.
Helius CEO Mert Mumtaz similarly urged users not to follow Drake's recommendation, arguing that careless fund movements could create a more immediate risk than an overnight cryptographic breakthrough.
A Crypto Problem or an Internet Problem?
The debate ultimately extends far beyond Bitcoin, Ethereum, or Solana. If AI genuinely discovers an efficient way to break widely used public-key cryptography, the consequences would reach banking, brokerage platforms, payment networks, TLS, software updates, and other critical internet infrastructure.
However, centralized companies and institutions could potentially respond faster to any threat. Blockchains face a tougher challenge because decentralized networks need coordination and consensus to make fundamental changes. That makes having a contingency plan before such a scenario emerges particularly important.
Solana already has one. Its Ed25519 design allows users to prove control of their underlying seeds and migrate to a safer signature scheme through a network upgrade, with a working proof of concept already available.
There is currently no public evidence that AI has broken ECDSA or another major cryptographic hardness assumption. In a recent tweet, Solana cofounder Anatoly Yakovenko described the panic as “fear based on uncertainty from smart people unable to predict the future,” but urged developers to ship post-quantum cryptography solutions.
For now, the argument centers on how seriously the industry should prepare for a possibility that researchers cannot rule out but also cannot demonstrate.
Read More on SolanaFloor
Securitize Launches Tokenized Stocks on Solana Ahead of 24/7 ICE Venue Listings
Jupiter Flips Kamino in Deposit Value to Become Solana’s Largest Lending Market
He Says The Next Cathie Wood Launches On Solana
