Ethereum researchers are urging crypto holders to begin a gradual, orderly migration of funds to fresh wallet addresses whose public keys have not been exposed, citing rapid advances in AI-driven mathematics that could weaken the elliptic curve digital signature algorithm (ECDSA) used to secure cryptocurrency wallets. The recommendation comes well before quantum computers are expected to pose a comparable threat, and it includes practical steps that everyday holders can take right away to lower their exposure.
What changed to trigger the warning
An Ethereum researcher raised the alarm in an October 8, 2026 social media post, pointing to a wave of newly published mathematical results from an AI lab as evidence that machine-aided mathematics is reaching into territory once thought to require years of human work. According to the post, the lab released hundreds of new mathematical findings across algebra, theoretical computer science, and mathematical logic on Tuesday, a pace that suggests AI is on track to tackle harder cryptographic problems sooner than expected.
The same lab had recently used a team of 10,000 autonomous AI agents working in parallel to solve the Navier-Stokes equations, one of the best-known open problems in mathematics and physics, in just 88 hours. That demonstration, the researcher wrote, was a sign that “long-held, unquestioned hypotheses have fallen,” and that elliptic curves in particular could be vulnerable to a future superintelligence because “curves carry rich structure, with room for fancy tricks like Schoof, Frobenius, pairings,” while hash functions are designed to minimize algebraic structure.
Why elliptic curve cryptography is the focus
ECDSA is the signature scheme that secures transactions on Bitcoin, Ethereum, and most other major blockchains. If an AI system, or any sufficiently capable attacker, could derive a private key from a public key in a realistic timeframe, every wallet whose public key has already been exposed would be at risk. Public keys are revealed the first time a wallet sends a transaction, which is why the recommended defense is to keep funds in addresses that have never been used to sign one.
The Ethereum co-founder publicly backed the call to take the threat seriously while cautioning against panic. “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 went further than the elliptic curve warning, noting that lattice-based cryptography, a family widely seen as the leading candidate for post-quantum security, could also be weakened by the same wave of AI-driven math advances. “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.” That broader concern is part of why Ethereum’s lean roadmap has been steering toward a “hash-only” cryptographic direction.
An investor and managing partner at a crypto-focused investment firm also endorsed the call, describing it as “a very sober call” and stressing that “the risk is not quantum, but just conventional mathematics overturning unproven cryptographic hardness assumptions.”
What a controlled migration looks like
The researchers framed the response as a “bunker mode” for crypto: a deliberate, staged move of assets to fresh addresses rather than a panicked sell-off. The researcher who first raised the issue laid out a clear sequence:
- Large and sophisticated crypto holders should be the first to migrate funds, since they are the most exposed and the best equipped to do so safely.
- Everyday users who find it easy to do so should consider keeping their funds in addresses that have not yet been used to make a transaction.
- After signing a transaction from any address, any remaining balance should be moved to a new address.
Both researchers warned that migration itself carries risk, and that speed is the wrong priority. “I personally have lost more money in botched migrations than I have lost in all hacks combined,” the Ethereum co-founder wrote, and the original post made the same point: “a rushed migration would do more harm than good.”
What this means for everyday crypto holders
For most users the practical message is straightforward. There is no need to panic-sell or rush funds into a new wallet today. There is also no reason to ignore the shift in the underlying math. Moving a long-term balance into a freshly generated address that has never broadcast a transaction is a low-cost step that meaningfully reduces exposure if ECDSA ever does become tractable for AI attackers. The bigger lesson from the past week of AI math results is that cryptographic assumptions once treated as “safe for decades” can move on much shorter timelines, and the wallet habits built today may matter well before quantum computers arrive.
FAQ
What is “bunker mode” in crypto?
Bunker mode is a term Ethereum researchers used to describe a controlled, gradual migration of crypto funds into fresh wallet addresses whose public keys have not yet been exposed, as a precaution against future cryptographic breakthroughs by AI systems.
How does AI threaten elliptic curve cryptography like ECDSA?
ECDSA is secured by mathematical problems that today are hard for conventional computers. Recent AI-driven math advances, including an AI-agent system that solved the Navier-Stokes equations in 88 hours, suggest that hard math problems are becoming tractable much faster than expected, raising the possibility that AI could one day derive private keys from exposed public keys.
Should crypto users move their funds to new wallets now?
The researchers urged a gradual, orderly migration rather than a rushed one. They recommend that large holders move first, that smaller holders keep funds in addresses that have never been used to send a transaction when feasible, and that any remaining balance be moved to a new address after each transaction. Both researchers warned that botched migrations have historically cost users more than hacks have.