# Ethereum researcher Justin Drake urges "bunker mode" after OpenAI math results revive fears for wallet signatures

Drake says the signature scheme that protects Bitcoin and Ethereum wallets could fail within months in a worst case, though no attack has been shown and about 6 million Bitcoin already sit at addresses whose public keys are visible on-chain.

- Published: 2026-10-08T05:50:08.479Z
- Canonical: https://polylog.news/crypto/2026-10-08/ethereum-researcher-justin-drake-urges-bunker-mode-after-ope
- Publisher: Polylog (Crypto desk)
- Section: crypto
- Sources: [CryptoSlate](https://cryptoslate.com/openais-math-breakthrough-raises-fears-ai-could-break-bitcoin-and-ethereum-security-within-months/), [Bitcoin Magazine](https://bitcoinmagazine.com/print/the-quantum-issue-bitcoin-quantum-exposure-at-block-950000), [Decrypt](https://decrypt.co/380363/ethereum-researcher-ai-break-encryption-before-quantum), [KuCoin News](https://www.kucoin.com/news/flash/ai-advances-in-math-spark-crypto-security-debate), [Crypto Briefing](https://cryptobriefing.com/justin-drake-bunker-mode-ai-ecdsa/), [KuCoin (Glassnode data)](https://www.kucoin.com/blog/in-measuring-bitcoin-s-quantum-exposed-supply-after-glassnode-s-2026-wake-up-call)

On October 7, Justin Drake, an Ethereum Foundation researcher and the author of the "Lean Ethereum" roadmap, told the crypto industry to start planning for what he called ["bunker mode"](https://cryptoslate.com/openais-math-breakthrough-raises-fears-ai-could-break-bitcoin-and-ethereum-security-within-months/). In bunker mode, holders would gradually move funds to addresses that have never signed a transaction, so their public keys stay hidden. "There is now reason to prepare for the possibility that ECDSA may fail before qday, with the worst-case scenario being months rather than years," Drake wrote, [according to Decrypt](https://decrypt.co/380363/ethereum-researcher-ai-break-encryption-before-quantum). ECDSA is the Elliptic Curve Digital Signature Algorithm, the method Bitcoin and Ethereum use to authorize transfers. "Q-day" is the industry term for the day a quantum computer becomes able to break that scheme.

The warning followed OpenAI's release of 722 mathematics papers on October 6. [KuCoin News reported](https://www.kucoin.com/news/flash/ai-advances-in-math-spark-crypto-security-debate) that the papers include counterexamples to several long-standing conjectures, among them work on the Erdős unit distance problem. Drake's argument rests on the mathematical structure of elliptic curves, not on any demonstrated attack. Elliptic curves have a rich mathematical structure, he said, and a system that produces new mathematics at scale could eventually find a way to crack the discrete-logarithm problem, the hard mathematical problem that currently keeps ECDSA secure. [Crypto Briefing reported](https://cryptobriefing.com/justin-drake-bunker-mode-ai-ecdsa/) that Ethereum co-founder Vitalik Buterin supported taking precautionary steps. Buterin also cautioned that the same AI capability could weaken post-quantum signature schemes as well, which would make migrating to them a weaker defense than the industry currently assumes.

No one has published an artificial intelligence (AI) attack on ECDSA. Drake described his call as preparation, not alarm. The exposure he is pointing to is measurable today. Glassnode data [summarized by KuCoin](https://www.kucoin.com/blog/in-measuring-bitcoin-s-quantum-exposed-supply-after-glassnode-s-2026-wake-up-call) puts 6.04 million BTC, or 30.2% of issued supply, at addresses whose public keys are already on-chain. About 1.92 million of those coins sit in early script formats that reveal the key by design. The other 4.12 million are exposed because owners reused addresses. Bitcoin Magazine's [Quantum Issue](https://bitcoinmagazine.com/print/the-quantum-issue-bitcoin-quantum-exposure-at-block-950000) sets out the same supply breakdown at block 950,000 and the practical problems of moving it.

The problem differs between the two largest chains. On Bitcoin, a fresh address hides the public key until the owner spends from it. On Ethereum, every externally owned account reveals its key with its first outgoing transaction, so a "fresh address" protects an Ethereum user only until that user's next transfer. Drake leads a roadmap that already plans to replace Ethereum's signatures with hash-based schemes, so an earlier deadline would add urgency to work he champions. Readers should weigh that interest alongside the technical substance of his argument.

## What this means

The mechanism here is a change in timeline, not in the underlying physics. Quantum risk depends on hardware whose progress outside observers can track. An algorithmic break, by contrast, would arrive as a published paper that anyone could use the day it appears, removing the multi-year warning period the market has assumed. The exposed parties are coins with visible public keys: early pay-to-public-key outputs, reused exchange and custodian addresses, and nearly every active Ethereum account. Exchanges and custodians that pool funds in a few reused addresses carry the most concentrated version of the risk. The consensus treats quantum risk as a distant tail that does not need pricing. The new element is a credible core researcher putting a "months" worst case on the record, which creates pressure on custodians to act even without proof of an attack. The likelier outcome over the next several weeks is procedural. Large custodians and wallet makers will publish address-rotation and key-hygiene guidance. Ethereum developers will cite the warning to justify faster work on signature agility, the ability to swap signature schemes without a full migration. Neither chain will see a forced migration. This read depends on no AI system publishing real progress on the elliptic-curve discrete-logarithm problem. A credible paper showing even a partial weakening of secp256k1, the specific curve both chains use, would prove it wrong and would turn guidance into an emergency migration.

## What to watch

- Whether OpenAI or other AI labs publish results that touch the elliptic-curve discrete-logarithm problem. That is the specific mathematics Drake's warning depends on, and one credible paper would change custody practice overnight.
- Whether major custodians and exchanges announce address-rotation programs for reused cold-wallet addresses. Their response will show whether the industry treats the warning as an operating risk or as commentary.
- How fast Bitcoin developers move proposals for quantum-resistant output types. Drake's framing makes those proposals relevant to a nearer-term threat as well.