Ethereum researchers warn AI could break wallet cryptography within months

By NewSunSEO | October 15, 2026

Fresh wallets, freshly funded and untouched by transactions, are emerging as the safest place for crypto holdings in the near term, as Ethereum researchers warn that the cryptography protecting today’s blockchains faces a faster threat from artificial intelligence than from quantum computers.

One Ethereum researcher urged the industry on Wednesday to prepare for “bunker mode,” saying AI could eventually make it possible to break the elliptic curve digital signature algorithm (ECDSA) used to secure cryptocurrency wallets. He pointed to a wave of AI-driven mathematical breakthroughs, including a recent demonstration in which OpenAI used a team of 10,000 autonomous AI agents working in parallel to solve the Navier-Stokes equation in just 88 hours, as evidence that the math behind wallet security is no longer moving at a human pace.

Why AI math now threatens wallet signatures

ECDSA, the scheme that proves ownership of funds on the Bitcoin and Ethereum networks, relies on the difficulty of problems built on elliptic curves. The researcher argued that those curves carry rich algebraic structure, room for techniques like Schoof’s algorithm, Frobenius endomorphisms, and pairings, which gives a sufficiently capable AI more to work with than hash functions, which are deliberately designed to minimize algebraic structure. As recent days have been humbling for human mathematical intuition, and long-held, unquestioned hypotheses have fallen, elliptic curves could be especially vulnerable to superintelligence.

The warning extends beyond just ECDSA. The Ethereum co-founder added lattice-based cryptography to the list of systems at risk, cautioning that schemes widely assumed to resist quantum attacks could also be weakened by AI-driven math. So far most people have been in the mode of thinking “elliptic curves broken, hashes safe, lattices safe,” but there is a good chance that the concrete security of lattices will take serious hits from the next two years of AI math, he wrote. This concern is a major reason why Ethereum’s lean roadmap has been going in the “hash-only” direction.

What bunker mode looks like in practice

The short-term recommendation is not a scramble. The Ethereum co-founder wrote that he does not recommend anyone rush to move funds to new wallets today, but the risks to cryptography from AI-accelerated math should be taken seriously. The first practical step is a controlled mass migration of assets to fresh addresses, with everyday users following once the large and sophisticated holders have paved the way.

The key idea is keeping funds in addresses whose public keys have not been revealed. On most blockchains, including Bitcoin and Ethereum, an address is derived from a hash, so the underlying public key stays hidden until the address sends its first transaction. Once a transaction is signed, the public key is broadcast to the network, which is the moment an attacker with the ability to break ECDSA would have something to work against. If it is not difficult for you, keeping your funds in addresses that have not yet been used to make a transaction is a good idea, the co-founder wrote.

After any transaction, the recommended habit is to sweep the remaining balance to a new, never-used address, so the next balance also sits behind a hidden public key.

Why a rushed migration can do more harm than good

Migrating funds is itself risky. The Ethereum co-founder disclosed that he personally has lost more money in botched migrations than he has lost in all hacks combined, a warning that human error during a panic can eclipse the threat users are trying to escape. The researcher echoed the point, noting that any migration should be gradual and carefully managed, warning that a rushed migration would do more harm than good.

Dragonfly managing partner Haseeb Qureshi described the call as a very sober one, framing the risk not as quantum but as conventional mathematics overturning unproven cryptographic hardness assumptions. The practical takeaway is to start small, rehearse the process with low-value amounts, verify every address character by character, and only scale up the migration once the workflow has proven safe.

How AI advances are changing the math timeline

OpenAI’s release of hundreds of new mathematical findings across algebra, theoretical computer science, and mathematical logic on Tuesday, on top of the 88-hour Navier-Stokes solution last month, was the catalyst for the warning. Each breakthrough shortens the perceived gap between today’s cryptography and an AI system capable of finding the structural weakness that would let it reverse private keys from signatures already exposed on chain.

The longer-term direction under discussion is moving Ethereum toward hash-based constructions, which offer less algebraic surface for an attacker to exploit. Until that work lands, the layered defense is the same as it has been for years, only with more urgency: keep balances in unused addresses, generate a fresh address for every new receipt, and avoid re-using addresses once their public keys are visible on chain.

FAQ

What does bunker mode mean for crypto holders?

Bunker mode is the idea of moving funds into fresh wallet addresses whose public keys have never been broadcast, so that any attacker capable of breaking ECDSA cannot target those addresses with already-exposed signatures.

How could AI break ECDSA?

By discovering new mathematical techniques that exploit the algebraic structure of elliptic curves, the same kind of AI-driven math research that recently produced new findings across algebra, theoretical computer science, and mathematical logic.

Is the threat from AI or quantum computers?

The researchers warn that conventional AI-accelerated mathematics could overturn unproven cryptographic hardness assumptions sooner than quantum computers would, and that even lattice-based schemes thought to resist quantum attacks could also be weakened.