Bitcoin and Ethereum face AI pressure: Holding positions at signed addresses enters "defense mode"

Ethereum researcher Justin Drake warns: AI may crack Bitcoin and Ethereum signatures earlier than quantum computers.

On October 7, 2026, Justin Drake, a researcher at the Ethereum Foundation, issued an article on the X platform pointing out that artificial intelligence (AI) may be the first to crack the signature algorithms of Bitcoin and Ethereum before quantum computers. He called on the industry to launch a "Bunker Mode" for large positions, in which assets are transferred in an orderly manner to addresses where a transfer transaction has never been signed.

The short answer to this outstanding question is: Your assets are not under acute threat today. There has been no actual attack demonstration against this signature scheme. Drake described the worst-case scenario, and the encryption community raised clear objections within hours. Still, this is crucial because it touches on a blockchain feature that most holders don't notice: the address only discloses its public key when the asset is first transferred out. By understanding this, you can reduce risks with ease, without panic, and without relying on new technology.

"Bunker Mode": Justin Drake's specific assertion on October 7

Drake proposed a specific yardstick to define what he considers to be a "hack." For him, once the private key can be calculated on existing hardware, such as a large graphics processor cluster, in a week or so, the scheme is considered broken. This is different from the theoretical vulnerability threshold that cryptographers have discussed for years and is set more conservatively.

He expressed the time frame as: In the worst case scenario, it could take months rather than years. As a trigger, he cited OpenAI's series of 722 mathematical results released on Tuesday. His reasoning as to why elliptic curves are particularly vulnerable is as follows: "Elliptic curves seem particularly vulnerable to superintelligence."

Drake announced that he will push Ethereum to "maximum defense acceleration" towards hash-based cryptography. At the same time, he issued a separate appeal to custodians, asking Binance, Bitbank, Robinhood, Bitfinex and Tether to strengthen their cold storage. The move is remarkable because these institutions hold large transaction balances and have been signing operations from their addresses for years.

ECDSA: The signature algorithm used by Bitcoin and Ethereum to authorize each transfer

ECDSA is fully known as the Elliptic Curve Digital Signature Algorithm. This is the scheme used by Bitcoin and Ethereum to verify whether the transfer comes from the authorized party. Each account contains two numbers: a private key that only you know, and a public key that can be deduced from the private key. This calculation process can only occur in one direction: the public key can be generated from the private key in an instant, while the reverse path is considered almost impossible under current technical conditions.

This unidirectionality is the foundation. If this mechanism breaks down, anyone who knows the address's public key can derive the private key from it and transfer funds. Neither a hardware wallet, mnemonic words, or an exchange with two-factor authentication can change this fact, because the attack occurs on the mathematics behind it, not on your device.

It is the following restriction that distinguishes panic from planning: Your Bitcoin address is not the public key itself, but its hash value. The hash is irreversible, and even a machine that can crack elliptic curves cannot calculate it backwards. As long as an address has never been transferred out, only that hash value is displayed on the chain. The public key is only exposed through a signature when you first send an asset from that address.

Why addresses become vulnerable only after the first transfer

A clear division can be drawn from this mechanism, where you can trace your own positions. Addresses where no asset has ever been sent are considered protected because their public key is hidden behind the hash; addresses where assets have been sent once are considered exposed. There is no transition state in between, and the boundary does not depend on how secure your wallet is, but on whether it has been signed.

For Ethereum, the situation is even more serious. The Ethereum address itself is derived from the public key, and every interaction with a smart contract involves a signature. The address of any user who has owned Ethereum for many years and has exchanged, pledged or authorized approval has been exposed. The situation for Bitcoin is more mixed: anyone who has sent assets to a new address and held them for a long time is in a relatively good position.

The actual dilemma is that consumption will inevitably lead to address exposure. Commonly used wallets cannot remain protected permanently. As a result, Drake's recommendations are mainly aimed at positions that are intended to remain in place for a long time, rather than funds used for day-to-day trading.

Satoshi Nakamoto's Shield: The first target is 20,000 old addresses holding 50 BTC

Drake makes an argument that has received little attention in the debate but is reassuring for small holders. It is speculated that about 20,000 exposed addresses belong to Bitcoin inventor Satoshi Nakamoto, each holding 50 BTC (a reward for early blocks). These addresses have not been moved for ten and a half years and are visible to everyone.

If an attacker can really crack ECDSA, he will face an order selection problem. Computing time is limited. Drake estimates that it takes about a week to crack each key on a large cluster, and at the top of the field are the 20,000 targets with the highest single address value. This effectively constitutes a kind of shield: anyone holding less than 50 BTC at an address will not be a primary target because the rewards of attacking other targets will be higher.

This shield is a time buffer rather than absolute safety. It assumes that attackers are economically rational, will not process all keys in parallel, and that no one will undermine the network for political purposes rather than self-interest. As a planning reference, it tells you that you have days to weeks to clean up, not hours.

Vitalik Buterin presses the brake: The cost of rushing action exceeds the sum of all hacking attacks

The most dramatic reaction came from Vitalik Buterin, whose views are divided into two parts. On substance, he agreed with Drake that AI-driven mathematical progress deserves more attention from the industry. But in terms of rhythm, he clearly opposed it. He said there was no time to rush the balance to a new wallet now.

His reasoning stems from his own experience and is the strongest argument against acting too quickly: According to him, the damage caused by operational errors is more than all the hacking attacks he has experienced combined. Typing the wrong destination, tampering with the clipboard with malware, mnemonics appearing in photos when setting up a new wallet: these mistakes often occur in haste, and unlike mathematical breakthroughs, they are real risks of the moment.

In addition, another risk emerges in every wave of migration. When headlines prompt holders to switch, guides, ancillary services and so-called inspection tools emerge, harvesting exactly what they promise to protect. Signing the authorization without reading it at this stage will result in funds being stolen by the "extractor" rather than by the superintelligence.

Cryptography community refutes: Coinbase's chief cryptographer believes there is no evidence that elliptic curve cryptography is about to collapse.

Coinbase's Yehuda Lindell, who is in charge of cryptography, is more acute than Buterin. He believes there is "absolutely no evidence" that the assumptions behind elliptic curve cryptography are close to breaking down. He pointed out that the mathematical advances made by AI systems this year do not pose a challenge to ECDSA because they involve different categories of problems.

Samson Mow, head of Jan3, also told readers to remain calm and take the warning lightly. Professional media unanimously classified the incident as a risk scenario: OpenAI has not yet proposed an actual attack on ECDSA, and Drake's proposed deadline of several months is an imaginary upper limit rather than a prediction.

It is worth noting that differences of opinion do not follow the previous division of camps. In March this year, after a widely watched paper was published by Google, Drake himself estimated that the probability of a quantum breakthrough by 2032 was 10% or more. He now believes AI is the faster path, which is a deepening of his own argument, not a reversal.

Hash-based signatures: WOTS and SPHINCS as targets of the Ethereum roadmap

Technical answers for this scenario have been around for years, namely hash-based signatures. Instead of relying on elliptic curve structures, it relies only on hash functions, which Drake considers relatively robust components. Buterin cites WOTS and SPHINCS as candidates and advocates avoiding lattice-based schemes when alternatives exist.

What does this mean for the schedule

Such switching is not a software update introduced overnight. It affects the signature format of every transfer, every wallet, every exchange, and every service that builds the transaction. For Bitcoin, such large-scale changes can only be achieved through consensus among developers, miners and industry, and this is the difficulty: proposing to invalidate old signatures after the deadline would be accused of depriving those who failed to move assets in a timely manner. The other side counters that if the old balance is not moved, it will become a hunting ground for the first successful attacker.

For you, this means that the protocol layer won't save you for months to come. What you have in your hands is the choice of address where you hold your position.

How does moving to a new address affect your holding cycle

At this point, the cryptographic debate turns into a tax issue, which is why many holders hesitate. The worry is: If I send coins to a new address, does the one-year holding period restart and lose the tax-exempt status of the proceeds?

The answer is reassuring. Transfers between two wallets that belong to you are not considered sales. Beneficial ownership has not changed, prices have not been realized, and no gains or losses have been incurred. Therefore, the holding period of your coin continues to run as if nothing had happened. The only exception is when you transfer assets through redemption (selling and re-buying): this constitutes a disposal with all tax consequences.

When moving between your wallets, what really matters is the documentation. The tax bureau saw the transfer on the blockchain, but could not determine whether it was a transfer between two owned wallets or sold to a stranger. Therefore, please keep track of how much money you transferred from which address to which address, and the date, and keep the original purchase receipt because it includes the date from which it was obtained. For large positions or nested transactions, the final decision is up to the tax adviser, rather than relying on the rules of thumb in the article.

Exchange escrow, hardware wallet or multi-signature: Where your public key is located

Whether you can take action depends on who holds your key. If your coins are deposited on an exchange, you don't own your address, but your claim to the institution. Whether its cold storage is located at an exposed address cannot be known from the outside, which is exactly what Drake is calling for. The choice left to you is to choose between trust and self-trusteeship.

Use your own wallet, the decision is in your hands. Every common hardware wallet can derive any number of addresses from a seed, one of which can remain untouched while you use the other for transactions. You don't need a second device or a new seed, just create a new account in your existing software.

The multi-signature construction deserves close scrutiny because multiple public keys are involved and the setting itself may expose them. People running such solutions should check which participation keys have been signed.

Old Bitcoin address: As long as there is no signature output, there is still plenty of time.

Today's conclusion is not sensational, but it is practical: the cracking of ECDSA has not yet occurred, has not been confirmed, and in the eyes of many cryptographers, it is unpredictable in the near future. What really happens is that this is an opportunity to look at the location of your money. This work takes half an hour and is useful for all future risks without rushing.

Check which addresses have sent assets:

Each blockchain browser displays outgoing transactions for the address. If you don't see any records there, your public key is hidden behind the hash. If you don't have any own address (because everything is on the exchange), the first step is to make a decision about the custody method.

Set up a separate account for the static portion:

Separate your inactive positions from the positions you trade and transfer them to an address that has never been signed with a cool check. Record dates, amounts and two addresses during this period; a portfolio tracker can help you do this.

Tell your exchange that you are paying attention to this:

If the majority of positions are on trading platforms, it is reasonable to question its custody practices, and such issues will be raised more frequently.

本周留下的箴言来自Buterin而非Drake：转移过程中最昂贵的部分，几乎从来不是它旨在防范的攻击。（截至2026年10月8日。本文不构成投资建议。价格和费用结构会发生变化；购买前请与提供商核实条款。）