Crypto Bunker Mode: Can AI Math Break Your Wallet Keys?
Crypto Bunker Mode: Can AI Math Break Your Wallet Keys?
For years, the long-term threat to Bitcoin and Ethereum wallets had one name: quantum computing. On 7 October 2026, during TOKEN2049 week in Singapore, Ethereum Foundation researcher Justin Drake named a second one. Writing on X, he called on the blockchain industry to “calmly start planning for ‘bunker mode'”, a gradual move of funds to addresses whose public keys have never been revealed, in case artificial intelligence finds a mathematical shortcut through the signatures that protect crypto wallets.
The trigger was a day earlier. On 6 October, OpenAI published 722 mathematical manuscripts produced by an unreleased internal model, claiming progress on hundreds of open problems. Drake’s argument is simple: if AI can now move research mathematics this fast, the assumption that elliptic-curve cryptography will stay unbroken until quantum computers arrive deserves a second look.
To be clear from the start: no practical attack on Bitcoin or Ethereum wallet keys has been demonstrated, and neither Drake nor Vitalik Buterin claimed one exists. This article explains what crypto bunker mode is, why serious researchers are discussing it, what an AI breakthrough could and could not do, and what sensible self-custody looks like while the debate plays out. Crypto remains highly volatile, and a security scare can move prices quickly whether or not the underlying threat ever materialises.
Key Numbers at a Glance
| Metric | Value |
|---|---|
| Bitcoin price | About $83,150 |
| Ether price | About $2,500 |
| Bitcoin all-time high | $126,080 (6 October 2025) |
| OpenAI math manuscripts released | 722, on 6 October 2026 |
| Manuscripts withdrawn by 7 October | 3, plus 14 revised |
| Bitcoin addresses with exposed keys on Project Eleven’s list | More than 14 million |
| Exposed early addresses Drake links to Satoshi | About 20,000, at 50 BTC each |
| Practical AI attack on wallet keys demonstrated | None |
Data as of 13 October 2026. Verify before publishing.
1. What Justin Drake Actually Proposed
Drake’s post was a call for preparation, not a claim of a break. His worst case is that AI could help find a way to recover a private key from a public key using ordinary computers, with no quantum machine needed, and that this could happen “in months not years”. In that scenario, he described breaking one elliptic-curve key as something that might take around a week on a large cluster of GPUs.
His recommendations were specific:
- Large holders move first, gradually and calmly, to fresh addresses whose public keys sit hidden behind a hash.
- “Sophisticated” holders lead, since they can migrate safely without rushed mistakes.
- Funds left in an address that has already signed a transaction should also move.
- Critical signers, such as oracles and layer 2 security councils, should consider rotating keys or adding hash-based signatures.
- Large custodians, which he named as Binance, Bitbank, Robinhood, Bitfinex and Tether, should consider stronger protection for cold storage.
He framed this as his personal view. Bunker mode is not an official Bitcoin or Ethereum emergency procedure, and current Ethereum guidance still says users need not act. Drake also said that leaving bunker mode safely would require “post-AI cryptography”, which in his view means security built on hash functions rather than on structured mathematics.
2. How Vitalik Buterin and the Critics Responded
Ethereum co-founder Vitalik Buterin agreed the risk is real but pushed back on urgency. “We should take the risks to cryptography from AI-accelerated math seriously,” he wrote, while adding: “I don’t recommend anyone scramble to move their funds to new wallets today.”
His most practical point was about the danger of the cure. “I personally have lost more money in botched migrations than I have lost in all hacks combined,” Buterin wrote. He did endorse one low-cost habit: keeping funds in addresses that have not yet been used to make a transaction.
Buterin then widened the debate. He warned that “there is a good chance that the concrete security of lattices will take serious hits from the next two years of AI math”. That matters because lattice-based schemes underpin ML-DSA, one of the post-quantum signature standards NIST finalised in 2024, and many in the industry had treated lattices as the safe destination after elliptic curves.
The pushback came quickly:
- Charles Hoskinson, the Cardano founder, argued that no credible attack on lattice problems has been identified after decades of cryptanalysis, and called Buterin’s suggestion of much larger key sizes “numerology” without a defined attack model.
- Samson Mow, CEO of Jan3, said there was no need to panic.
3. AI vs Quantum: Why Crypto Bunker Mode Is a Different Kind of Threat
Every Bitcoin and Ethereum wallet is protected by the Elliptic Curve Digital Signature Algorithm (ECDSA) on a curve called secp256k1 (Bitcoin also supports Schnorr signatures on the same curve). Your public key is derived from your private key by a one-way calculation. Running it forwards is instant; running it backwards, the “discrete logarithm problem”, is believed to need on the order of 2^128 operations with the best known classical methods. That is far beyond any computer on Earth.
The quantum threat, which we covered in our earlier Bitcoin quantum computing risk analysis, is different in kind. Shor’s algorithm solves the discrete logarithm problem efficiently, but only on a large, error-corrected quantum computer that does not yet exist. The threat has a visible hardware roadmap, so the industry can watch it coming.
An AI-driven threat would not need new hardware at all. It would be a new piece of mathematics, a shortcut nobody has found, that could run on existing GPUs once discovered. That is why Drake leans on hash functions: elliptic curves have rich algebraic structure, and structure is exactly what a pattern-finding system could exploit, while hash functions are designed to have as little structure as possible.

Why OpenAI’s math release set this off
The OpenAI release is the evidence Drake pointed to. It is real, but it is also messy:
- 722 manuscripts, posted to GitHub on 6 October, from a model OpenAI has not named or released.
- Claimed progress on 372 open problems across geometry, algebra, computer science and other fields.
- About half released unconfirmed, according to an OpenAI spokesperson, with only part of the set checked in the Lean proof assistant.
- Errors already found: on 7 October OpenAI withdrew three manuscripts over a sign error and revised 14 others.
Mathematicians were split. Cornell’s Alex Townsend said OpenAI “should have announced the manuscripts that were lean verified first”, while MIT’s Andrew Sutherland called the quick corrections “the responsible thing to do”. The point for crypto holders is not any single proof. It is the pace. For a broader view of how quickly frontier models are improving, see our look at GPT-6 Astra’s capabilities and cost.
Quantum vs AI at a glance
| Feature | Quantum threat | AI math threat |
|---|---|---|
| What it needs | New, large quantum hardware | A new mathematical idea, then ordinary GPUs |
| Visibility | Hardware progress is public and gradual | A breakthrough could arrive without warning |
| What it targets | ECDSA, Schnorr and RSA via Shor’s algorithm | Whatever has exploitable structure, possibly including lattices |
| Current status | No machine capable of it exists | No practical attack demonstrated |
4. What AI Could and Could Not Break
Even in Drake’s worst case, an AI shortcut would not “collapse” Bitcoin or Ethereum. Mining and block hashes rely on SHA-256 and Keccak, which are hash functions, not elliptic curves. The ledger’s history would stay intact. What would be at risk is the ability to spend coins from specific addresses.
The key detail is exposure. Most modern addresses publish only a hash of the public key, and the public key itself appears on-chain only when the owner spends from that address. A key-recovery attack needs the public key, so:
- Exposed: very early Bitcoin outputs that pay directly to a public key, Taproot outputs (which carry a key from the start), any Bitcoin address that has been reused after spending, and on Ethereum any account that has ever sent a transaction.
- Hidden: a fresh hash-based Bitcoin address, such as a standard SegWit address, that has received coins but never spent from them.
- Scale: Project Eleven’s Bitcoin Risq List tracks more than 14 million addresses that hold funds and have exposed public keys.
Ethereum is more exposed than Bitcoin by design, because accounts are reused for every transaction. Stablecoins and tokenised funds on Ethereum are controlled by the same signature system. Our Ethereum 2026 outlook covers the roadmap that is now moving towards hash-based cryptography.

The Satoshi factor
Drake pointed to roughly 20,000 exposed early addresses he links to Satoshi Nakamoto, each holding 50 BTC. If a shortcut were ever found, those coins would be the obvious first target, and he described them as a “Satoshi’s shield” that gives smaller wallets some early warning. The market consequence is the real concern here: coins that have not moved in more than 15 years suddenly being spent would be a supply shock, and very likely a price shock, long before most individual holders were touched.
5. Market Reaction: Fear, Bonds or Both?
Bitcoin did fall while the debate ran. It slid to about $82,300 during Asian hours on 8 October, roughly 4% below its high near $86,600 two days earlier, and as of 13 October trades around $83,150.
It would be too simple to blame bunker mode. CoinDesk’s market coverage tied the slide mainly to bond markets, with the 30-year US Treasury yield at 5.71% and Fed minutes showing most officials expect another rate hike, and described derivatives activity as de-risking rather than fresh bearish bets. The security debate added to an already nervous tape; it did not create it.
6. Practical Risk Management for Crypto Holders
Whatever you think of Drake’s timeline, the steps he proposes are mostly good habits anyway. The two incidents that actually cost holders money this year had nothing to do with AI math: a firmware flaw in Coldcard devices, covered in our Coldcard hack explainer, and tampered Ledger devices in October.
On 9 October, wallets were drained that had been set up on Ledger devices sold through CryptoBilis, a reseller in Southeast Asia. Ledger said all confirmed cases involve that reseller and that its own systems and direct sales were not compromised. It advised anyone who bought from that reseller in the past 90 days not to set up the device, and anyone already using one to move funds to a new device with a newly generated recovery phrase.
A sensible checklist, in order of effort:
- Stop reusing addresses. Receive to a fresh address each time; most modern wallets do this by default.
- Know which of your addresses are exposed. On Bitcoin, any address you have spent from has revealed its key. Moving the remainder to a fresh address is low-cost hygiene, not panic.
- Migrate slowly if you migrate. Test with a small amount first. Buterin’s warning about botched migrations is the bigger day-to-day risk.
- Buy hardware wallets from the maker or an authorised channel, and treat any device that arrives with a pre-filled recovery phrase as compromised.
- Split large holdings across more than one wallet. After the Coldcard exploit, Binance founder Changpeng Zhao noted that “even hardware wallets can have bugs” and suggested diversifying, while acknowledging that splitting has its own risks. In October he added: “I am not against hardware wallets, or self-custody. Different wallets/product for different use cases.”
For the mistakes holders make most often, see our guide to Bitcoin cold wallet myths.

7. Bull vs Bear Case
š Bull Case
- No practical attack on secp256k1 exists, and the best known classical attack has not improved in decades.
- Exposure is manageable: coins in unspent, fresh Bitcoin addresses keep their public keys hidden.
- Ethereum’s roadmap is already moving towards hash-based signatures, and the debate may speed up that work.
- A clear early warning signal exists, since Satoshi’s exposed coins would likely be targeted before ordinary wallets.
š» Bear Case
- More than 14 million Bitcoin addresses already have exposed public keys, and every active Ethereum account has one.
- A mathematical breakthrough, unlike quantum hardware, could arrive without a visible roadmap.
- Buterin’s lattice warning, if right, weakens the post-quantum standards the industry planned to migrate to.
- A rushed migration, or Satoshi’s coins moving, could trigger a sharp sell-off whatever the technical reality.
8. Outlook: What Would Change the Picture
Most likely, bunker mode stays a planning exercise. Drake himself said existing cryptography has not been broken, and the most useful outcome is that custodians, oracles and protocol teams start treating key exposure as a real cost.
What to watch:
- Peer-reviewed or Lean-verified results touching elliptic-curve discrete logarithms. A confirmed result there would change everything; a press release would not.
- Custodian responses, especially whether the large exchanges Drake named rotate cold-storage keys.
- Ethereum and Bitcoin signature upgrades, including hash-based options.
- Any movement of early-era coins, which would be the clearest real-world alarm.
Follow OneMoreBitcoin on YouTube for regular crypto market updates.
FAQ
What is crypto bunker mode?
Crypto bunker mode is Justin Drake’s term for a planned, gradual move of funds to fresh addresses whose public keys have never been revealed on-chain. It is a precaution against AI possibly finding a shortcut through elliptic-curve signatures, not a response to any break that has happened.
Can AI break Bitcoin encryption today?
No. No practical attack on Bitcoin or Ethereum wallet keys has been demonstrated, and neither Drake nor Buterin claimed one. The debate is about how fast AI is accelerating mathematics and whether the industry should prepare earlier than its quantum timeline assumed.
Should I move my Bitcoin to a new address now?
Buterin’s advice was not to scramble. Keeping funds in addresses that have never been used to send a transaction is good practice, and moving leftover funds from an address you have spent from is reasonable hygiene, but test with a small amount first because botched migrations are a bigger everyday risk.
What is an exposed public key?
It is a public key that is visible on the blockchain. Most addresses only show a hash of the key until you spend from them; after that, the key is public. Any key-recovery attack, quantum or AI-driven, needs that public key to work.
Is Ethereum more at risk than Bitcoin?
In terms of exposure, yes. An Ethereum account reveals its public key with its first transaction and is normally reused, while a Bitcoin user can receive to a fresh, never-spent address each time. Both chains use the same secp256k1 curve.
Conclusion
Crypto bunker mode matters because it shifts the security conversation from a hardware threat everyone can see coming to a mathematical one that might not announce itself. The opportunity is that the fix is mostly good hygiene already within every holder’s control. The biggest risk is less the mathematics than the reaction to it: panicked migrations, custodian mistakes, or a supply shock if Satoshi-era coins ever move.
What would change the thesis is a verified result against elliptic-curve discrete logarithms. Until then, treat this as a reason to tighten self-custody, not a reason to abandon it.
Disclaimer:
This article is for educational and informational purposes only. It does not constitute investment, financial, legal or tax advice, and it is not a recommendation to buy, sell or hold any digital asset.
Cryptocurrency is highly volatile and speculative. Prices can move sharply at any hour of any day, and you can lose some or all of the capital you invest. Digital assets are not bank deposits, they are generally not covered by deposit protection or investor compensation schemes, and the regulatory treatment of a token, platform or service can change quickly. Further risks include exchange or custodian failure, smart contract vulnerabilities, bridge and oracle failure, loss of private keys, and scheduled token unlocks that expand supply.
Investors should conduct their own due diligence and consider their own circumstances, time horizon and risk tolerance before making any financial decision. Consult a licensed financial adviser if you are unsure. Past performance is not indicative of future results.
OneMoreBitcoin and OneMoreMoney.com accept no responsibility for any loss incurred from reliance on the information provided in this article.