Ethereum researcher Justin Drake has urged the cryptocurrency industry to brace for a scenario in which artificial intelligence breaks the signature algorithms protecting assets on networks such as Bitcoin and Ethereum within months, well before the expected arrival of quantum computers.
AI Threat Raises Urgency for Crypto Security
Drake cautioned that recent advancements in AI could soon allow attackers to crack the private keys behind wallets holding bitcoin, ether, and other tokens. He recommended that large holders begin moving their assets to addresses whose public keys have never been exposed onchain in order to mitigate potential risks.
No practical attack has been demonstrated so far. However, Drake emphasized proactive measures to shield substantial funds and called for the industry to calmly prepare for what he referred to as “bunker mode.”
Drake wrote on X, “Today I call upon the blockchain industry to calmly start planning for ‘bunker mode’.”
The mathematics securing crypto wallets relies on elliptic curve cryptography, where deriving a private key from a public key is intended to require infeasible computational resources. Drake warned that powerful AI models might soon provide a shortcut, enabling attackers to reveal private keys using conventional hardware.
Millions of bitcoin are currently held in addresses with exposed public keys. On Ethereum, any account that has ever made a transaction has revealed its public key, potentially increasing vulnerability if AI-assisted attacks become viable in practice.
OpenAI Developments and Recent Attacks
Drake’s warning followed OpenAI’s release of 722 mathematical manuscripts generated by an unreleased AI model. These documents included solutions to complex problems, such as the Navier–Stokes problem, and signaled the rising capabilities of AI to solve highly specialized mathematical challenges. Some results were quickly verified by external researchers.
Recent incidents have demonstrated that AI-powered systems can already identify vulnerabilities in smart contract code and wallet software. For instance, Anthropic’s research showed that frontier models could generate working exploits against DeFi contracts. In July, the Bitcoin Red Team leveraged AI to scan nearly 400 Bitcoin software projects, identifying about 5,000 possible bugs, 85 of which were considered critical.
In one high-profile incident, Coldcard hardware wallets were compromised through a firmware vulnerability, resulting in a theft of at least 1,367 BTC. Coinkite, the manufacturer, indicated that AI-assisted analysis may have uncovered the flaw. Additionally, attacks on BTCPay Server and Core Lightning infrastructure have been linked to vulnerabilities identified through AI-generated bug reports.
Researchers have also used AI coding agents to improve calculations relevant to quantum attacks on cryptographic systems, though these efforts still require quantum hardware and remain theoretical in nature.
Mini dictionary: Elliptic curve cryptography — A public-key cryptography approach widely used in blockchain wallets, designed to enable secure digital signatures by relying on the mathematical difficulty of reversing the relationship between a public key and its corresponding private key.
Vitalik Buterin: Don’t Rush Migrations
Vitalik Buterin, co-founder of Ethereum, acknowledged the potential risks AI poses to cryptographic security and noted that some quantum-resistant alternatives, such as lattice-based cryptography, may also be at risk. He stated that while hash-based signatures offer fewer opportunities for shortcuts, they are not completely immune to novel attack vectors.
Buterin highlighted, “I personally have lost more money in botched migrations than I have lost in all hacks combined,” cautioning users against moving funds in haste.
Buterin advised that users should only reduce public key exposure where practical and avoid unnecessary, rushed migration processes, emphasizing that operational errors can result in significant losses. The Ethereum Foundation currently targets December 2029 to upgrade the network with quantum-resistant cryptography, though Drake warned that vulnerabilities may materialize much sooner if AI advances rapidly.
Sophisticated users are advised to consider moving funds to addresses where public keys remain undisclosed, as this reduces the attack surface for possible AI-assisted exploits. Meanwhile, researchers and developers continue to evaluate long-term cryptographic upgrades for both resilience to quantum computing and now, increasingly, AI-powered threats.
