Justin Drake, an Ethereum Foundation researcher, is calling on the crypto industry to prepare for what he calls “bunker mode,” a controlled migration of funds to new wallet addresses whose public keys stay hidden behind a hash. His argument, posted on X on October 7 with a note viewed close to 4 million times, is that advances in AI could give attackers a way to break the elliptic-curve signatures securing bitcoin and ether wallets before powerful quantum computers arrive.
Drake pointed to a recent release of over 700 mathematical results across algebra, number theory and topology by OpenAI, verified with the Lean proof assistant, as evidence of how fast machines are moving through mathematics. “Long-held, unquestioned hypotheses have fallen,” he wrote. In his worst case, Drake said a breakthrough could arrive “within months not years,” and defined a break as a private key recovered within roughly a week on hardware like a large GPU cluster. His warning focuses on ECDSA, the signature scheme used across most major networks, and hinges on one technical detail: an attacker needs the public key, which is usually exposed on-chain when a wallet spends funds.
What bunker mode means in practice
Under Drake’s proposal, large holders would move first, transferring coins to addresses that have never signed a transaction, so their public keys remain unrevealed. Holders would then be encouraged to move remaining funds again after any signing activity, repeating the cycle as they transact. He stressed the migration should happen gradually, and that “a rushed migration would do more harm than good.” No such break has been demonstrated, and the proposal sits somewhere between a research wishlist and a permanent migration plan.
Ethereum co-founder Vitalik Buterin pushed back on urgency. “I do think the risk from AI-accelerated math should be taken seriously, but holders should not be scrambling to move funds,” he said. Samson Mow, CEO of bitcoin technology firm Jan3, wrote that there was no need to panic “because an Ethereum researcher is saying silly things.”
Why elliptic curves, and why now
Modern cryptography leans on elliptic curves because no known classical algorithm breaks them efficiently. Quantum computers running Shor’s algorithm would, on a large enough machine. But Drake argues AI could produce a targeted mathematical shortcut before those machines exist, a separate pathway that would let conventional hardware attack exposed public keys. In earlier posts he has favoured hash-based cryptography, which he says has less algebraic structure for AI to exploit, and has urged Ethereum to accelerate its planned post-quantum upgrades. “We already have a Post Quantum migration to execute, let’s not add more confusion,” he wrote on X this week.
The practical effect of his proposal would be significant for large holders and custodians, including exchanges, whose wallets hold broad key visibility across chains. Cold wallets that have never signed a transaction would already qualify as safe under the plan. Hot wallets, which sign constantly, would not, and would need a rotation strategy that remains workable under real trading volumes.
Market reaction was muted. Bitcoin traded near $83,000 as of Thursday morning, roughly where it has sat since Tuesday, with no evidence the post itself moved prices. A few tokens describing their cryptography as quantum-resistant, including Algorand, which uses post-quantum Falcon signatures for its state proofs, and Starknet, whose proofs use hash-based cryptography, gained in a quiet market. Algorand led a flat CoinDesk 100 with a roughly 9 percent gain on the day.
Not the same as a quantum warning
Discussions around post-quantum risk in crypto have run for years, usually tied to timelines for quantum hardware that most researchers put a decade or more out. Drake’s post shifts the argument: it treats AI as a route that could arrive sooner, and frames preparation as a process that belongs in the present rather than a migration once a threat is confirmed. Whether that framing changes how exchanges or funds manage key rotation is unclear, and the Ethereum Foundation has not made an official statement alongside the post.
Drake’s address of the issue goes beyond a single tweet. He has written on Ethereum’s planned upgrade path, its hash-based signature roadmap, and bridge opportunities between research groups and client teams. Critics of his framing point out that any AI attack on elliptic curves faces the same mathematics that has resisted attack for decades, and that a general improvement in reasoning does not automatically produce a new cryptanalytic breakthrough. Drake has not claimed otherwise, describing the post as a planning recommendation rather than a warning of an active vulnerability.
Sources: CoinDesk Markets Today; The Defiant; Cryptobriefing; Cryptopolitan; U.Today.
The wider conversation on quantum risk
Drake’s warning lands at a moment when several networks are already moving to post-quantum cryptography. Algorand uses Falcon signatures for state proofs, Starknet has built its proofs on hash-based schemes, and research groups within Ethereum have been drafting a migration plan for years. What changes in Drake’s framing is the timeline. A quantum threat with a decade of lead time supports a slow, coordinated upgrade. A mathematical shortcut that could arrive within months would compress that schedule and leave exposed wallets with little margin.
Custodians have started to say publicly that they are tracking the issue. A handful of exchanges, including Coinbase and Kraken, have published notes on post-quantum readiness in the past two years, though none has rewritten its wallet infrastructure around the concern. Hardware wallet makers face similar choices, since a wallet that signs often accumulates public key exposure over time.
No regulator has commented on the post, and there is no indication that a coordinated migration by large holders is under way. Drake has said the argument is not that such an attack exists today, but that the worth of a preparation plan only shows once it is too late to build one. Whether that case convinces exchange operators and custodians to change how they manage keys is likely to be tested over the coming months, rather than in a single week.