
Breakdown: Are cryptocurrencies entering "bunker mode"? Can AI really crack wallets?
Author: A Fox
Compiled by: TechFlow
TechFlow Overview: Ethereum Foundation researcher Justin Drake recently issued a warning that AI's explosive progress in advanced mathematics could crack the underlying cryptography of crypto wallets before quantum computers do, calling on the entire industry to enter asset defense "Bunker Mode." Vitalik Buterin also weighed in with a follow-up post. Is this mere alarmist speculation born of unwarranted concern, or an imminent security crisis? This article meticulously deconstructs the one-way mathematical mechanism of Elliptic Curve Cryptography (ECC), the additional shield provided by hash-hidden public keys, the truth behind Anthropic's use of Claude to solve the 25-year-old NIST random seed puzzle, and the technical roadmap ranging from post-quantum signatures to emergency rollback mechanisms. In the face of the AI wave, the crypto world needs neither panic nor haste; instead, it should methodically advance the upgrade of its underlying cryptographic defenses.

Ethereum Foundation researcher Justin Drake issued a warning this month that artificial intelligence may soon breach the underlying mathematical defenses protecting crypto wallets, calling on the entire industry to enter "Bunker Mode."
Nearly all cryptocurrencies and DeFi protocols rely heavily on this mathematical framework, which is also true for nearly half of the internet, though no substantial cryptographic breaches have occurred thus far.
In fact, the news has been positive as AI initially tackled an unsolved mystery in elliptic curves, demonstrating that long-suspected backdoors planted by the US National Security Agency (NSA) simply do not exist.
Corresponding defense schemes are being rapidly developed. Transitioning to post-quantum signature technologies can simultaneously mitigate risks from both AI and quantum computing, and the industry has already formulated emergency contingency plans for potential crises.
A few weeks ago, Ethereum researcher Justin Drake posted on X, noting that as AI evolves at a breakneck pace, it could soon gain the capability to crack the critical cryptography used in blockchain. Vitalik quickly echoed these sentiments.
Both stated that we should prepare adequately for this moment. Drake advocates entering "Bunker Mode," where everyone takes extra precautions to protect digital assets. However, they both later clarified that there is no need for panic! This seemingly contradictory series of statements prompted me to delve deeper into the topic today and thoroughly unpack its nuances.
Bunker Mode
On October 7, Ethereum Foundation researcher Justin Drake released a warning tweet that sent shockwaves through the crypto industry. He pointed out that AI may soon breach the underlying cryptography protecting crypto wallets, potentially even beating quantum computers to the punch.

Figure 1: Justin Drake posts on X warning that AI breakthroughs may threaten encryption cryptography before quantum computer threats materialize.
What truly spurred Drake was AI's explosive progress in mathematics. Just the day prior, OpenAI had announced hundreds of new mathematical breakthroughs its AI achieved in algebra and theoretical computer science; a month earlier, reports indicated that an OpenAI cluster comprising 10,000 AI agents successfully solved a famous physics puzzle in just 88 hours.
Drake named his response strategy "Bunker Mode." Like people seeking shelter in underground bunkers before a storm, the core idea suggests cryptocurrency holders transfer assets to more secure locations and wait patiently until the industry deploys a more robust next-generation cryptographic defense system.
Operationally, this so-called "shelter" refers to a brand-new wallet address that has never initiated any outgoing transfers, with its corresponding public key never exposed on the public chain (the deeper cryptographic logic behind this will be detailed below).
Drake calls for a planned, gradual migration of funds to these newly unexposed addresses, advising users to move remaining balances to new addresses immediately after each outgoing transaction from old ones. However, he explicitly warns against rushing it, specifically noting that "the harm caused by hasty, panicked migrations will far outweigh potential gains."
Vitalik Buterin subsequently posted on X supporting Drake's view. He stated he does "not recommend hastily and blindly migrating all assets to new wallets at this moment, but we must take seriously the potential threat AI-driven mathematical breakthroughs pose to underlying cryptography."

Figure 2: Vitalik Buterin posts in support of Drake's warning, urging the industry to face the potential cryptographic risks head-on.
Other top scholars within the industry reacted with much greater composure. Yehuda Lindell, Chief Cryptographer at Coinbase, bluntly called Drake's remarks "a very poor framing," emphasizing that modern cryptography, tested over decades, currently shows zero evidence of fatal flaws.
Meanwhile, Lucien Bourdon, an analyst at hardware wallet manufacturer Trezor, warned that widespread panic-driven asset migration across the network could easily become a feast for phishing scammers.
The core cryptography Drake and Vitalik worry about being breached is the renowned Elliptic Curve Cryptography (ECC), which serves as the bedrock for securing user private keys and controlling asset access!
Elliptic Curves
Every crypto wallet holds a pair of keys: a private key, a massive secret number used to manage and transfer your on-chain funds; and a public key, generated directly from the private key via elliptic curve mathematical formulas.
The security foundation of the entire modern encryption system rests entirely on the robust property of this mathematical calculation allowing only "one-way traffic": deriving the public key from the private key is computationally trivial, while mathematically reversing the process to crack the private key from the public key is extremely difficult (requiring billions of years at current human computational capacities).
Every time you initiate an on-chain transaction, you are digitally signing it with your private key. Anyone on the network can quickly verify the signature's authenticity against your public key, with absolutely no chance of exposing your private key during verification. However, if someone mathematically discovers a shortcut for reverse inference, they could instantly crack the corresponding private key using only the exposed public key, then freely sign transfer transactions and drain any target wallet's assets!
DeFi smart contracts like Aave do not hold private keys themselves, but almost every critical component surrounding them relies on private keys to operate:
• Your personal wallet: Signs every deposit and withdrawal transaction.
• Protocol admin private keys and multi-signature contracts: Responsible for core protocol upgrades.
• Cross-chain bridge validator nodes: Sign confirmation credentials for cross-chain messages.
This scope of impact extends far beyond the crypto circle itself. Bank online banking systems, end-to-end encrypted messaging software, and the secure transmission of most of the modern internet all rely on the same foundational elliptic curve cryptography for protection.
Nevertheless, your wallet address actually possesses a natural built-in defensive armor: because a wallet address is not the public key itself, but a derived result generated by applying a one-way cryptographic "hash" function to the public key, and any math-based cracking techniques targeting elliptic curves are powerless against hash functions.
Your actual public key is only broadcast and exposed on the public chain when you actively send your first outgoing transaction from that address. Therefore, funds stored in a new address that has "never initiated any outgoing transactions" remain safely shielded by the indestructible hash function. This is the fundamental cryptographic logic behind Drake's recommendation to disperse assets into brand-new, clean addresses.
This mirrors the exact same potential vulnerability discussed in my in-depth essay on quantum computers earlier this year: ancient Bitcoin addresses where public keys are fully exposed on-chain (including Satoshi Nakamoto's dormant wallets) will face the highest potential threat in the future.
NIST Random Seeds
In 2000, the US National Institute of Standards and Technology (NIST), responsible for setting technical standards, published a series of elliptic curve parameters. These standards later became among the most widely applied cryptographic curves globally. However, these curves were generated using mysterious "random seeds" provided by the US National Security Agency (NSA), and officials provided no reasonable explanation for the specific origins of these seed numbers at the time.
For 25 years, the global cryptographic community remained skeptical: Did the NSA secretly embed some unknown decryption backdoor when selecting these random seeds? Countless researchers conducted specialized investigations, and the community even established high bounties seeking answers to how these seeds were generated.
On October 16 this year, renowned AI firm Anthropic announced that its Claude model completely solved this century-long mystery haunting the industry for a quarter-century in less than half an hour!

Figure 5: Anthropic announces that its Claude model cracked the NIST elliptic curve random seed mystery plaguing the industry for 25 years in under half an hour.
It turned out that the origin of these seeds was far from the sinister conspiracy previously imagined. They were merely standard outputs from pseudo-random number generators commonly used on computers in the 1990s. Their initial generation values directly stemmed from the clocks of those computers, with timestamps accurately tracing back to 1997 and 1998. Today, any developer on the network can flawlessly reproduce these exact seeds without deviation!

Figure 6: Cryptography expert Steve Weis publishes an analysis explaining the generative source of NIST random seeds, confirming they are merely outputs of standard pseudo-random algorithms.
Consequently, Anthropic reached its final conclusion: "This major discovery reinforces our confidence that these standardized curves were not deliberately implanted with any hidden backdoors or engineered security flaws from the outset."

Figure 7: Diagram illustrating the logical derivation of NIST curve random seeds.
In reality, Bitcoin and Ethereum wallets utilize a different elliptic curve named secp256k1, whose underlying parameters never harbored any mystery regarding secret seeds from the beginning. Therefore, this deciphering discovery concerning NIST seeds does not directly impact the everyday security of your wallet.
Nevertheless, this marks the first time in human history that AI successfully attacked an unsolved elliptic curve mystery that baffled cryptographers for decades. Its ultimate outcome not only failed to destroy cryptography but rather proved the innocence and reliability of existing cryptographic systems. Cutting-edge AI tools, which attackers might exploit in the future, are currently on the frontlines helping defenders fortify the walls of this digital world.
Solutions
Addressing the risk of potential decryption attacks on elliptic curve cryptography, the industry has long prepared ready-made countermeasures. This solution set represents the defensive fortifications the entire crypto sector has been advancing for years against quantum computer threats.
Post-quantum safe signatures completely abandon the elliptic curve mathematical framework. Therefore, a smooth transition to post-quantum cryptography can thoroughly eliminate systemic risks along this dimension, regardless of whether the entity that first cracks elliptic curves is a quantum computer, cutting-edge AI, or top human mathematicians. This underlying cryptographic upgrade can permanently neutralize all potential threats!
NIST officially finalized the first batch of post-quantum signature algorithm standards as early as 2024, which I deeply analyzed in a prior research report on quantum computers. One of these standards is built entirely upon hash functions, which serve as the most battle-tested cornerstone of modern cryptography.
Vitalik has also cautiously pointed out that some post-quantum encryption schemes heavily rely on complex mathematical structures known as "Lattices." He publicly warned: "The rapid breakthroughs AI will make in mathematics over the next two years have a significant probability of dealing a heavy blow to the practical security of lattice-based cryptography."
In contrast, pure hash-based signature algorithms rely on neither lattice mathematics nor elliptic curves. This is precisely the core reason why Ethereum's technical roadmap is firmly heading toward the "Hash-only" architecture advocated by Vitalik.
Leading the charge on the forefront of this defensive front, ironically, is the very whistleblower himself, Justin Drake. In the "Lean Ethereum" development roadmap he first unveiled in July 2025, he explicitly drives Ethereum's overall evolution toward a hash-based signature system; the Ethereum Foundation also formally assembled a dedicated post-quantum security R&D team in January this year.

Figure 8: The post-quantum roadmap for Lean Ethereum, led by Justin Drake.
Vitalik's long-term planning objective is to establish a comprehensive post-quantum protective net for the entire Ethereum ecosystem before the end of 2029. Thus, the researcher sounding the security alarm is simultaneously paving the path to lead the industry out of trouble!
The Bitcoin ecosystem is also taking action, albeit at a more cautious pace. In February this year, an improvement proposal named BIP-360 was formally added to the official Bitcoin Improvement Proposal directory, laying the architectural foundation for introducing quantum-resistant secure addresses. Although the proposal remains in draft stage and requires broad network consensus before formal activation, the relevant code is currently running stably on test networks.

Figure 9: Official homepage for Bitcoin Improvement Proposal BIP-360, aimed at introducing post-quantum secure address formats.
Faced with the upgrade of cryptographic defenses, the crypto industry is certainly not fighting alone. Mainstream tech products like Google Chrome, Apple iMessage, and Signal have already begun switching to quantum-resistant encryption algorithms, and the entire internet industry possesses tremendous motivation to jointly solidify this underlying defense line.
As early as 2024, Vitalik prospectively envisioned Ethereum's ultimate contingency plan in the event of an extreme, sudden cryptographic breach:
The entire network would urgently execute a state rollback at the last possible moment before widespread asset theft occurs, completely suspending ordinary transaction functionality. Subsequently, users could safely recover and migrate their assets to brand-new quantum-resistant wallets by providing mnemonic phrase signature proofs that map to their private keys.
This emergency mechanism works because private keys were originally derived from mnemonic phrases via hash function calculations, with elliptic curve cryptography only playing a role later when transforming the private key into a public key.
Therefore, even if confronted with an irresistible sudden extreme event, the vast majority of users' on-chain assets can still be preserved under an orderly contingency plan!
Of course, this belongs to an extreme last resort necessitated by dire circumstances, which would cause severe disruption to normal blockchain network operations. Therefore, the ideal path is to allow all mainstream public chains to complete smooth hot-patch upgrades well before elliptic curves are truly and substantially compromised.
Stay Calm
Facing this series of intricate industry discussions, what exactly should you, as a ordinary holder, do right now? For the vast majority: basically, nothing.
If it feels seamless and effortless to you, cultivating the habit of storing long-dormant assets in clean addresses that have never initiated outgoing transfers is undoubtedly a highly sound preventive measure, a practice Vitalik himself praises. But there is absolutely no need to panic and hastily shuffle assets around, as blind asset migration inherently carries enormous operational risks. As Vitalik shares from personal experience: "The financial losses I've suffered from various fumbling, erroneous asset migrations exceed the sum of all hacker theft losses I've experienced over the years."
Additionally, you need to be highly vigilant regarding any unfamiliar private messages or scam scripts urging your emergency transfer of crypto assets under the guise of "asset safety." This is precisely the go-to predatory tactic phishing gangs employ during trending news cycles.
My personal judgment is: this is indeed a long-term systemic risk warranting proactive preparation and serious attention from the entire industry, and the sector is going all-out to advance defensive preparations. To date, no existing cryptographic system has faced substantial decryption, countermeasures are clear and comprehensive, the scholars who sounded the initial alarm are precisely the core forces crafting the antidote, and even in case of emergencies, the industry has ready its ultimate safety contingencies.
This aligns perfectly with my previous conclusions drawn in research reports on quantum computers: as the spear attacking cryptography grows sharper, the shield defending cryptography evolves synchronously, and AI undoubtedly empowers both sides in the game of offense and defense.
The crypto industry has always cherished the fine tradition of facing the most severe threats head-on under the spotlight and collaboratively solving problems openly in plain sight. And this ongoing dialectic and mobilization regarding AI and cryptography is precisely the new chapter in this resilient story of evolution!













