Ethereum researcher Justin Drake argues that AI-powered mathematical advancements could pose risks to current cryptographic systems sooner than expected, urging the cryptocurrency sector to begin planning for what he calls a “shelter mode” defense mechanism.
Drake said that large and institutional investors, in particular, should consider moving their assets in a controlled manner to new addresses where the public keys are not yet visible on-chain. He noted that these addresses should not have previously signed any transactions, adding that it might be safer to transfer remaining assets to a new address once another address has signed a transaction.
However, Drake emphasized that investors should not rush or panic. He stated that moving assets to addresses whose public keys have not yet been revealed is a simple and preventative step that can be taken with the existing wallet infrastructure.
According to an Ethereum researcher, the industry should now seriously consider the possibility that the elliptic curve-based digital signature algorithm known as ECDSA could be compromised before the widely anticipated quantum breaking event dubbed “Q-Day.”
Drake suggested that, in a worst-case scenario, ECDSA private keys could become crackable within months using current hardware. Drake’s definition of “cracking” encompasses obtaining a private key in about a week using currently available hardware such as a large GPU cluster.
In his statement, Drake also drew attention to some recent developments in the field of mathematics. He noted that some mathematical assumptions long considered true have recently been disproven, and argued that artificial intelligence systems could significantly increase the speed of mathematical discovery.
“It is said that decades can happen in weeks. We may be entering a period where centuries of mathematical progress are accomplished in a few weeks.”
Drake argued that elliptic curve cryptography, particularly due to its structural properties, may be more vulnerable to advanced artificial intelligence systems. He also reminded us that hashing functions are designed to contain as few algebraic structures as possible.
The researcher also pointed out that efficient algorithms developed for quantum computers can, in some cases, later inspire algorithms that run on classical computers. Therefore, he argued that theoretically, new methods could emerge that could break RSA using elliptic curve cryptography, similar to the Shor algorithm, with classical computers.
Drake stated that large crypto companies and institutional custody services should lead the sector in defense. He noted that companies like Binance, Bitbank, Robinhood, Bitfinex, and Tether could strengthen their cold wallet structures, and emphasized the need to particularly monitor Bitcoin addresses where the public key has become visible on the blockchain.
However, according to Drake, small Bitcoin investors may have relatively more time in some cases. The researcher suggested that thousands of open addresses, believed to belong to Satoshi Nakamoto and each containing approximately 50 BTC, could be a prime target for potential attackers, describing this as “Satoshi’s shield.”
Drake stated that critical infrastructure that requires continuous transaction signing, such as Oracle systems and Layer 2 network security councils, should consider regularly exchanging their ECDSA public keys, and that additional layers of security can be created with hash function-based signature systems like SPHINCS.
Drake believes that in the long term, the crypto industry needs to prepare for the “post-AI cryptography” era. He states that he prefers summary function-based cryptography and argues that structured mathematical assumptions such as elliptic curves, lattice-based systems, or isogenies should be avoided as much as possible.
Drake noted that Ethereum’s roadmap already includes a transition to hash function-based cryptography and end-to-end formal verification to counter quantum computing threats, but said that recent AI developments need to reassess and accelerate the current timeline.
*This is not investment advice.
