OpenAI math breakthroughs raise ‘bunker mode’ alarm from Bitcoin researcher Justin Drake

OpenAI’s latest advances in mathematics have prompted a leading Ethereum researcher to warn that crypto’s core wallet security could face an unexpected AI threat.

On Oct. 7, Ethereum researcher Justin Drake urged the blockchain industry to prepare for what he described as “bunker mode,” including a controlled migration of digital assets into fresh addresses whose public keys have never been exposed.

Drake said large and sophisticated holders should consider moving most of their assets to addresses that have never signed transactions. Once those addresses are used to send funds, he recommended transferring any remaining balance into another fresh address.

His warning followed OpenAI’s Oct. 6 release of a broad collection of new mathematical results produced by an internal frontier model.

The company said it tested the model across thousands of problems, published many of the resulting proofs with computer-checkable Lean formalizations and spent the equivalent of about three hours of ChatGPT Pro reasoning on the average result.

Drake argued that the accelerating pace of AI-driven mathematical discovery should force the crypto industry to reconsider assumptions about how long elliptic-curve cryptography will remain secure.

“In the worst case,” he said, an effective break of the Elliptic Curve Digital Signature Algorithm, or ECDSA, could arrive “in months, not years.”

OpenAI’s announcement does not report a practical attack on ECDSA or RSA. Drake’s months timeline is a worst-case conjecture, not a demonstrated capability.

Drake warns AI could shorten crypto’s security timetable

Bitcoin and Ethereum rely heavily on elliptic-curve cryptography to establish ownership and authorize transactions.

Standard Ethereum externally owned accounts use ECDSA on the secp256k1 curve. Once an account sends a transaction, its public key can be reconstructed from information placed onchain. Someone capable of efficiently deriving the corresponding private key could then take control of the assets.

Standard Ethereum accounts whose public keys remain unexposed have an additional layer of protection: only the address, a hash of the public key, is visible, Ethereum’s documentation says.

The industry has traditionally treated that problem primarily as a future quantum-computing risk.

Ethereum has already established a dedicated post-quantum security effort, with work underway on hash-based signatures, account abstraction and replacements for other cryptographic systems vulnerable to sufficiently powerful quantum computers. Core post-quantum infrastructure is currently targeted for roughly 2029, although the roadmap remains subject to change.

Drake is challenging the assumption that quantum computing will necessarily be the first technology capable of breaking those systems.

He pointed to recent surprises in mathematics and argued that sufficiently capable AI could uncover a classical algorithm that dramatically reduces the difficulty of recovering private keys.

That would change the industry's timetable because it would remove the need to wait for large fault-tolerant quantum computers.

Drake cited the precedent of quantum algorithms occasionally inspiring faster classical approaches and said researchers should remain open to a classical counterpart to Shor’s algorithm, which would threaten both elliptic-curve cryptography and RSA.

Whether such an algorithm exists remains unknown.

Europol separately added urgency to the broader issue, warning that quantum computing could eventually undermine cryptography protecting cryptocurrency wallets and urging the industry to begin preparing before the threat becomes practical. The agency said uncertainty over timing should not delay migration because upgrading systems and coordinating defenses could itself take years.

Crypto holders could start moving before cryptography changes

Drake’s proposed response does not require waiting for new blockchain infrastructure.

Drake proposed moving funds to addresses that keep public keys hidden behind hashes and avoiding unnecessary outgoing transactions. That protection depends on the address type: Bitcoin Taproot outputs expose a public key from the outset. Taproot also uses Schnorr signatures rather than ECDSA, although both rely on the secp256k1 curve.

Drake urged large custodians to lead that transition, specifically naming Binance, Bitbank, Robinhood, Bitfinex and Tether as firms with an opportunity to harden cold-storage practices.

He also recommended more aggressive precautions for critical blockchain infrastructure. Oracles and layer-2 security councils, for example, could rotate ECDSA keys after signing messages or combine existing signatures with hash-based systems such as SPHINCS.

Those measures would amount to an interim defense rather than a permanent solution.

Drake said his preferred long-term approach would rely heavily on hash-based cryptography, which has less algebraic structure for future AI systems to exploit than elliptic curves, lattices or isogenies.

Ethereum is already moving partly in that direction. Its post-quantum roadmap includes hash-based validator signatures and mechanisms designed to let individual accounts eventually adopt different signature schemes without requiring the entire network to migrate at once.

Drake cautioned against a rushed migration, warning that hurried transfers could create more risk than they remove. But he said Ethereum’s existing timelines should now be revisited as AI changes the assumptions underlying them.

Ethereum’s second annual post-quantum research retreat is scheduled for Oct. 9 to Oct. 12, while Drake said he plans to address institutional participants in London next month as he pushes for faster defensive preparations.