Recent advances in artificial intelligence in mathematics have reignited the debate over cryptocurrency security. Justin Drake, a researcher at the Ethereum Foundation, believes that advances in AI could accelerate the discovery of vulnerabilities in the cryptography used by Bitcoin and Ethereum, and recommends that major holders consider taking preventive measures. However, his warning is not based on any known attack against the algorithm in question: several specialists, including Vitalik Buterin and Ledger’s chief technology officer, urge people to take the risk seriously without giving in to panic or making hasty transfers.
The caution recommended by some industry players is rooted in an essential question: could advances in AI undermine the mathematical mechanisms that protect digital wallets? For now, there is no evidence that the signature systems used by major blockchains can be broken by artificial intelligence. But the accelerating pace of mathematical discoveries is fueling concern and prompting some in the industry to examine precautionary measures.
On October 7, 2026, Justin Drake urged blockchain players to calmly prepare for a scenario involving strengthened defenses, which he describes as “bunker mode.” His most immediate proposal is to move significant funds to new addresses that have never been used to sign a transaction. However, he stresses one point: a poorly organized migration could cause more losses than it would prevent.
A concern that goes beyond the quantum threat alone
Discussions about cryptocurrency security often focus on the possible arrival of quantum computers capable of weakening certain cryptographic systems. This milestone, nicknamed “Q-Day,” remains uncertain. Drake’s warning concerns another scenario: AI tools could help researchers or attackers discover new mathematical methods more quickly, without waiting for quantum computers to become sufficiently powerful.
According to the researcher, recent advances in AI in mathematics justify revisiting certain assumptions that were previously considered solid. In an extreme scenario, he raises the possibility of recovering private keys in a relatively short time from exposed public keys. This is a risk hypothesis, not a description of an attack that can already be carried out.
Why the ECDSA algorithm is at the center of the debate
Bitcoin and Ethereum rely, among other things, on ECDSA, a digital signature system based on elliptic curves. When a user authorizes a transaction, a signature allows the network to verify that it was approved by the holder of the corresponding key, without that person disclosing their private key.
The security of the process depends on the asymmetry between two operations. A private key can be used to calculate the associated public key. By contrast, recovering the private key from the public key is considered impractical with the methods and computing resources available today. This mathematical difficulty protects funds from people seeking to reproduce the owner’s signature.
Justin Drake believes elliptic curves could be particularly interesting for an AI capable of exploring new mathematical avenues, because they have a rich structure. He envisages that a previously unknown method could make private key recovery much faster. But no public demonstration currently shows that the ECDSA used by cryptocurrencies can be broken in this way.
An exposed public key does not mean a wallet has already been compromised
The distinction between a public key and a private key is important. The former can be used to verify signatures or receive funds; the latter gives control over them. A public key revealed on the blockchain therefore does not, by itself, provide access to assets. The danger raised in the debate would depend on the discovery of a new method capable of deriving the private key from the public key.
The details also vary depending on the address type and blockchain. In some Bitcoin formats, the public key only appears when funds are spent. On Ethereum, a transaction signature can make it possible to recover the public key associated with the address. These differences explain why some specialists advise limiting address reuse, while emphasizing that this precaution is not proof that an attack is imminent.
Why some investors are considering new addresses
Justin Drake’s advice is to transfer significant holdings to addresses that have never signed a transaction. In formats where the public key has not been revealed, this can reduce exposure to a hypothetical attack targeting public keys that are already visible. For an investor, the aim would therefore be to limit the information that could be exploited if a new cryptographic vulnerability were discovered.
This measure does not make funds invulnerable to every threat. It does not protect against the theft of a recovery phrase, malware, sending funds to the wrong address, fraud, or a platform breach. It is intended only to reduce a category of risk related to public key exposure and a possible break in ECDSA.
Drake initially urges major holders and organizations with specialized resources to examine this possibility. He also calls on platforms and companies that hold significant reserves offline to assess their storage procedures and migration plans. In his analysis, very large wallets could attract more attention from attackers if a method for recovering keys became practical.
The operational risk of a rushed migration
Moving cryptocurrency requires great care. An error in the destination address, fee management, network selection, or device verification can lead to the irreversible loss of funds. A migration carried out under pressure from a warning can also expose an investor to fake websites, fraudulent applications, and deceptive requests for a recovery phrase.
That is why Drake himself recommends not rushing. Vitalik Buterin, Ethereum’s co-founder, also stresses the dangers of a poorly configured upgrade. He points out that failed migrations can be costly and that users should not make improvised transfers simply because a threat hypothesis has been raised publicly.
Caution therefore means staying informed, checking the recommendations of your wallet or platform, and never sharing your private keys. Any transfer procedure should be understood and tested with a level of care appropriate to the value held. A social media post is no substitute for technical expertise or the official guidance of a wallet provider.
Mathematical advances fueling the warning
Drake’s position came after OpenAI published hundreds of mathematical papers produced with the help of an artificial intelligence model. Some of the results challenged conjectures or intuitions that had been established for a long time. For the researcher, these advances indicate that AI systems could help produce mathematical discoveries at an unexpected pace.
This development does not mean that AI can already break cryptocurrencies. The ability to solve difficult mathematical problems and the ability to actually recover a private key are two different things. A relevant method would still need to be devised and verified, and then shown to work at scale on accessible hardware. It is precisely the gap between a theoretical discovery and an operational attack that fuels disagreement among specialists.
Nevertheless, recently published mathematical research has led several players to reconsider how long current security assumptions will remain valid. Drake’s warning is thus part of a long-term discussion: rather than waiting for proof of a break, some believe it is useful to identify the most exposed assets and potential alternatives now.
A warning disputed by cybersecurity experts
Not all specialists share Justin Drake’s assessment. Charles Guillemet, Ledger’s chief technology officer, acknowledges that AI’s mathematical performance is impressive, but considers a conventional break of ECDSA within a few months to be an excessively alarmist hypothesis. In his view, a practical attack of this magnitude would have consequences far beyond the cryptocurrency sector.
Indeed, ECDSA and other public-key cryptography mechanisms are used in many digital systems. A method capable of broadly compromising these protections would also threaten banking services, secure connections, software signing, and other infrastructure. Such a break would therefore affect not only Bitcoin or Ethereum: it would require a much broader response across the entire technology ecosystem.
Guillemet nevertheless distinguishes the value of good security practices from the idea of a general migration. For major holders, avoiding address reuse after a transaction can be a reasonable precaution. By contrast, triggering mass transfers without a demonstrated threat could multiply errors and create confusion, without any guarantee of improving overall security.
Vitalik Buterin urges people to take the risk seriously without panicking
Vitalik Buterin also responded to Drake’s comments. He believes AI could weaken certain cryptographic assumptions sooner than expected, but says there is currently no reason for a hasty response. He urges holders to distinguish between preparing for a possible scenario and the existence of a vulnerability that is actually being exploited.
Buterin also raises the need to consider so-called post-AI and post-quantum cryptography. Hash-based schemes are among the approaches he considers promising for certain signatures, because they rely on mathematical structures that differ from those of elliptic curves. However, this is not a universal solution: requirements, performance, and applications vary.
Owen Simonin, known by the pseudonym Hasheur, also pointed out that Drake’s statements describe a hypothesis, not a confirmed vulnerability. This distinction is central for investors: a warning may justify monitoring and preparation, but it does not prove that a wallet is currently exposed to a new attack.
How holders can respond judiciously
For investors, the first step is to reinforce essential security practices that are already important: keep your recovery phrase offline, never share it with anyone, verify addresses before sending funds, and use reputable wallets. These precautions remain relevant regardless of the debate about AI, because they address the most common risks, such as phishing, malware, and user error.
Major holders can examine how their addresses expose public keys, the consequences of any reuse, and the transfer procedures provided by their custody solution. Companies and platforms, for their part, should document their response plans, test their storage systems, and avoid relying on improvised procedures.
It is also useful to distinguish technical warnings from commercial solicitations. A message claiming to offer an urgent migration, asking for a recovery phrase, or directing you to an unknown wallet should be treated with extreme caution. Since blockchain transactions are generally irreversible, a fake security instruction can itself become a way to steal funds.
On a different note, readers who would like to explore other published profiles can read this one about Jadeline Leroux-Fleury, an animal lover.
The transition to more resilient cryptography is being prepared
The issue goes beyond asset transfers. Blockchain networks will need to assess how to evolve their signatures and protocols if current cryptographic assumptions prove insufficient. A transition of this kind takes time: solutions must be designed, analyzed, tested, and then adopted by wallets, platforms, and users.
Hash-based cryptography is one of the approaches being studied for certain digital signatures. Other families of methods, particularly those based on lattices, are also being considered in post-quantum research. Each approach, however, involves its own trade-offs. A method may offer interesting guarantees in one area while being unsuitable for other uses, such as public-key encryption or complex proofs.
Discussions about AI are therefore encouraging development teams to revisit their migration timelines and security assumptions. They also serve as a reminder that blockchain security does not depend solely on its algorithms: wallet software, key management, user practices, and governance mechanisms all play a role in protecting funds.
A possible threat, but no demonstrated break
As of October 12, 2026, discussions among researchers and cybersecurity leaders reflect disagreement about the timing and likelihood of an attack against ECDSA. Justin Drake believes it is prudent to anticipate rapid advances in AI’s mathematical capabilities. Charles Guillemet considers a conventional compromise in the near term insufficiently substantiated to justify a general migration. Vitalik Buterin, for his part, recommends measured vigilance and technical preparation without impulsive moves.
This disagreement explains why the warning is prompting a response across the industry. Investors need to keep two things in mind: AI’s mathematical capabilities are advancing and deserve serious attention, but there is no public evidence that the keys to Bitcoin or Ethereum wallets can currently be recovered by AI. Further technical assessments, audits, and decisions by development teams will help clarify the actual level of risk and the appropriate responses.
