A senior researcher at the Ethereum Foundation has issued a pointed warning to the broader cryptocurrency community: artificial intelligence, not quantum computing, may be the first technology capable of breaking the cryptographic signature schemes that secure billions of dollars in digital assets. Justin Drake made the case publicly, urging holders to take measured, deliberate action now rather than waiting for a crisis to materialize.
Drake's recommendation is straightforward but carries serious implications. He advised Ethereum holders to migrate funds to unused, fresh wallet addresses — a practice that limits exposure by ensuring that a wallet's public key has never been broadcast on-chain. When a transaction is sent from an address, the corresponding public key becomes visible on the blockchain, creating a potential attack surface. Addresses that have never transacted keep the public key hidden, providing an additional layer of protection against any future cryptanalytic breakthrough, whether driven by AI or quantum hardware.
What makes Drake's warning notable is not the destination — the eventual threat to elliptic curve cryptography has been discussed in the blockchain space for years — but the proposed timeline and the proposed agent. The conventional narrative inside and outside crypto has positioned quantum computers as the looming existential threat to public-key cryptography. Governments, standards bodies including the U.S. National Institute of Standards and Technology, and blockchain developers have been steadily working on post-quantum cryptographic standards under the assumption that sufficiently powerful quantum machines are still years, possibly decades, away from posing a practical threat.
Drake is challenging that sequencing. His argument centers on the possibility that AI systems capable of discovering novel mathematical techniques could accelerate cryptanalysis in ways that are far harder to predict than the physical engineering constraints on quantum hardware. Quantum computing progress is measurable in qubit counts, error correction rates, and coherence times — all observable, trackable metrics. An AI-driven mathematical breakthrough is, by contrast, discontinuous and unpredictable. A sufficiently capable model trained on the landscape of number theory and algebraic geometry could, in principle, surface insights that human mathematicians have missed, compressing what was expected to be a multi-decade threat timeline into something far shorter.
This is not a fringe concern. The intersection of machine learning and formal mathematics has become one of the more active and surprising areas of AI research. Large language models and specialized AI systems have already demonstrated non-trivial performance on mathematical reasoning benchmarks, and research groups are actively exploring AI-assisted theorem proving. The distance between "AI helps mathematicians work faster" and "AI discovers a tractable attack on elliptic curve discrete logarithm" is uncertain — but it is not obviously infinite. Drake's point is that the community should not assume the quantum timeline gives it a comfortable buffer if a parallel threat vector is developing on an entirely different, less legible schedule.
The practical guidance — move to fresh addresses calmly, without panic — is deliberately measured in tone. Drake is not predicting imminent collapse of Ethereum's security model. He is making a probabilistic argument: given the uncertainty around AI's mathematical capabilities over the next several years, the cost of taking a precautionary step is low, and the potential benefit is meaningful. Keeping funds on addresses whose public keys have already been exposed is an unnecessary risk that holders can eliminate with a single transaction today.
The broader infrastructure implication deserves attention. Ethereum's roadmap already incorporates long-term thinking about cryptographic agility — the ability to swap out cryptographic primitives as the threat environment evolves. Account abstraction, which has been a consistent theme in Ethereum's development, opens the door to wallets that could enforce post-quantum signature schemes at the smart contract level without requiring a hard fork. But those solutions are still maturing, and adoption will not be instantaneous even once the tooling is production-ready. In the interim, behavioral hygiene — like Drake's advice to use fresh addresses — is the most accessible mitigation available to ordinary holders.
What this means for the market and for developers is a sharpening of urgency around cryptographic resilience that does not depend on quantum milestones. Teams building wallet infrastructure, key management systems, and custody solutions should treat Drake's warning as a prompt to audit their exposure to reused addresses and to accelerate integration of post-quantum signature research into their roadmaps. The threat model has always been "when, not if." The uncomfortable addition to that model is now "and we may not see it coming from the direction we expected."
Written by the editorial team — independent journalism powered by Bitcoin News.