Ethereum co-founder Vitalik Buterin has backed a warning from one of the network's best-known researchers that artificial intelligence could undermine the cryptography securing hundreds of billions of dollars in cryptocurrency before quantum computers ever get the chance.
The warning, from Ethereum researcher Justin Drake, urges the industry to prepare for what he calls "bunker mode": an orderly, gradual migration of funds to fresh wallet addresses whose public keys are not exposed on-chain. For more context on this story, see our ongoing more AI stories.
The Warning, and Who Is Behind It
Writing on Wednesday, Drake argued that recent progress in AI-driven mathematics means the elliptic curve digital signature algorithm (ECDSA) — the mathematical foundation securing Bitcoin and Ethereum wallets — can no longer be assumed safe on a decade's horizon.
ECDSA relies on the assumed difficulty of the elliptic curve discrete logarithm problem: it is computationally infeasible to derive a wallet's private key from its public key using today's computers and known algorithms. Every transaction a wallet signs publishes its public key to the blockchain, which is why the structure of that mathematics — not just the secrecy of the key itself — is what stands between an attacker and the funds.
His trigger was OpenAI's release of hundreds of new mathematical findings this week across algebra, theoretical computer science and mathematical logic, which Drake cited as evidence of how quickly AI is advancing in pure mathematics. The company has previous form here: last month it used a team of 10,000 autonomous AI agents working in parallel to produce a solution to the Navier-Stokes equations, one of the most famous unsolved problems in mathematics and physics, in just 88 hours.
"Recent days have been humbling for human mathematical intuition. Long-held, unquestioned hypotheses have fallen," Drake wrote, arguing that elliptic curves are especially vulnerable because they "carry rich structure, with room for fancy tricks like Schoof, Frobenius, pairings." Hash functions, by contrast, "are designed to minimize algebraic structure."
What Buterin Actually Said
Buterin's endorsement was measured rather than alarmist. "I don't recommend anyone scramble to move their funds to new wallets today," he wrote. "But we should take the risks to cryptography from AI-accelerated math seriously."
He then extended the concern beyond elliptic curves to lattice-based cryptography — the family of schemes widely assumed to resist quantum attacks. "So far most people have been in the mode of thinking 'elliptic curves broken, hashes safe, lattices safe,'" Buterin wrote. "But there is a good chance that the concrete security of lattices will take serious hits from the next two years of AI math."
That concern, he said, is a major reason Ethereum's lean roadmap has been moving in a "hash-only" direction — relying on cryptographic primitives with minimal algebraic structure that AI-assisted mathematics would find hardest to attack.
The Industry Reacts
Dragonfly managing partner Haseeb Qureshi endorsed taking precautions, describing Drake's warning as "a very sober call." "The risk is not quantum, but just conventional mathematics overturning unproven cryptographic hardness assumptions," he wrote on X.
That distinction is the crux of the debate. Post-quantum cryptography has a roadmap: the US National Institute of Standards and Technology has standardized lattice-based replacements, and blockchains have years-long migration plans. But those replacements rest on hardness assumptions that are conjectured, not proven. If AI-accelerated mathematics starts toppling long-standing conjectures — as the past week's cascade of AI-derived results suggests it can — the quantum-safe consensus itself becomes shakier.
The Practical Advice: Fresh Addresses
Alongside the long-term cryptographic questions, both researchers emphasized a concrete mitigation available today: keeping funds in addresses where the public key has never been revealed on-chain. A wallet's public key becomes exposed the moment it signs a transaction, which is why Drake recommends a gradual migration to fresh addresses rather than a panicked exodus.
For exchanges, custodians and institutional holders, the operational takeaway is unglamorous but concrete: key hygiene, address rotation and a preference for hash-based constructions belong on the security roadmap now, not after a mathematical breakthrough makes headlines.
The timing question is deliberately unsensational. Neither Drake nor Buterin put a date on when — or whether — AI-assisted mathematics could produce a practical attack on elliptic curve signatures. What changed this week is the pace of the underlying field: proofs and results that would once have taken research teams years are now appearing in machine-generated batches, and cryptographers are re-examining which of their hardness assumptions were ever actually proven.
Neither researcher claimed a break is imminent. The point both emphasized is that the assumption chain underneath modern cryptography is only as strong as the mathematics it rests on — and that mathematics is now being rewritten at machine speed.
As AI systems move from assisting human proofs to generating original mathematical results at scale, the security teams managing trillions in digital assets may need to start treating frontier AI labs' publications the same way they treat zero-day disclosures: as early warnings.
For more on how frontier AI research is reshaping adjacent industries, follow AI Buzz Wire's AI business section.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
