Ethereum researcher Justin Drake has urged the cryptocurrency industry to prepare for what he calls “bunker mode,” warning that artificial intelligence could, in the worst case, help break the mathematical protections securing Bitcoin and Ethereum wallets within months rather than years.
No practical attack on the wallet signatures of either network has been demonstrated. However, Drake has advised large holders to begin moving funds gradually to fresh addresses whose public keys have never appeared onchain. Ethereum co-founder Vitalik Buterin agreed that the risk deserves attention, while cautioning that hurried migrations could create losses of their own.
Why wallet keys are at the centre of the warning
A crypto wallet uses a private key—a secret number—to authorise transactions. Its corresponding public key allows the network to verify that a payment was approved by the rightful owner. Producing the public key from the private key is straightforward, but reversing that process is designed to require an impractical amount of computing power.
Drake’s concern is that sufficiently capable AI could discover mathematical shortcuts that make this reversal possible on ordinary computers, without requiring the quantum machines that have traditionally been viewed as the main long-term threat to today’s cryptography.
The potential exposure would extend beyond coins held in individual wallets. Millions of bitcoin are stored at addresses whose public keys have already been revealed onchain. On Ethereum, an account exposes its key when it sends a transaction. The same signature system also controls stablecoins and tokenised funds issued on the network.
A move to an address whose public key has not yet been exposed could remove the starting information an attacker would need. Drake therefore recommended that sophisticated or large holders take preventive action over time rather than wait for a demonstrated attack.
AI capabilities are already affecting crypto security
The warning followed OpenAI’s release of 722 mathematical manuscripts produced by an unreleased model tested on about 4,000 research problems. OpenAI said some of the work had computer-checkable proofs, while other results remained unverified and could contain errors. The company also said the model had solved the Navier–Stokes problem, one of seven Millennium Prize mathematics challenges, in a result announced the previous month.
Each manuscript used, on average, computing resources equivalent to roughly three hours of ChatGPT Pro reasoning, according to OpenAI. An outside researcher subsequently reran the computer check on one result concerning the speed at which computers can multiply large grids of numbers and found that it held.
Crypto-related AI attacks have already produced concrete security concerns, even though none has demonstrated a break of Bitcoin or Ethereum wallet keys. Anthropic researchers reported last December that leading AI models could write working exploits against simulated versions of real decentralised-finance contracts. In July, the Bitcoin Red Team used AI models to examine 390 Bitcoin software projects in about 27 hours, identifying nearly 5,000 possible flaws, including 85 rated critical.
On July 30, an attacker drained at least 1,367 BTC from Coldcard hardware wallets by exploiting a five-year-old firmware vulnerability. Coinkite, the maker of the wallets, said it suspected AI had helped identify the flaw. BTCPay Server later confirmed thefts from merchants’ Lightning nodes through a vulnerability first identified during an AI-assisted audit. Core Lightning developers issued an emergency warning on Aug. 27 after AI-generated reports revealed real software weaknesses.
Quantum-resistant plans face a new question
Ethereum’s foundation has set December 2029 as its target for moving the network to quantum-resistant cryptography. Drake’s warning presents a more immediate worst-case timeline, because it concerns a possible classical attack assisted by AI rather than an attack requiring quantum hardware.
Buterin said some proposed quantum-resistant systems could also be affected by progress in AI-driven mathematics. Lattice-based cryptography, used in a digital-signature standard approved by the US National Institute of Standards and Technology, relies on mathematical problems considered difficult for both conventional and quantum computers. Buterin wrote that AI could make significant advances against the concrete security of lattice systems over the next two years.
Ethereum’s proposed long-term redesign increasingly favours hash-based signatures. These use digital fingerprints intended to be difficult to reverse, and Buterin sees fewer opportunities for unexpected mathematical shortcuts, although he acknowledged that no design is guaranteed to be immune to attacks.
Buterin supported reducing public-key exposure where practical, but opposed rushing into migrations. He said mistakes during previous migrations had cost him more money than all hacks combined. The competing concerns underline the difficulty of “bunker mode”: delaying action could leave exposed keys vulnerable if Drake’s worst case materialises, while moving funds quickly can itself lead to errors and losses.
Moneyiar market picture
Moneyiar data at 21:33–21:34 on Oct. 8, 2026, showed Bitcoin at $81,622.80, down 1.97% over 24 hours and 3.45% over seven days. Ethereum was at $2,472.71, down 3.68% in 24 hours and 8.17% over seven days.
Other tracked assets were also lower: Shiba Inu stood at $0.0000052947, down 2.01% over 24 hours and 8.03% over seven days, while Uniswap traded at $7.3319, down 5.38% and 18.89% over those periods, respectively.
The security warning was reported by CoinDesk. It describes a potential technological risk, not a confirmed compromise of Bitcoin or Ethereum wallets.
