A leading researcher at the Ethereum Foundation is urging the blockchain community to brace for the possibility that artificial intelligence, rather than quantum computing, might be the first technology to breach the cryptographic safeguards protecting prominent cryptocurrencies.
Justin Drake, a well-known figure within Ethereum’s research sphere, expressed his concerns on X this Wednesday, advocating for a “controlled mass migration” of digital assets to new wallet addresses. He describes this approach as entering “bunker mode,” prompted by what he perceives as a growing threat: AI-enhanced mathematical advancements that could compromise the elliptic curve digital signature algorithm (ECDSA), which Bitcoin and Ethereum rely on for transaction validation.
Drake’s advisory marks a significant shift in the longstanding discourse around cryptographic vulnerabilities. Traditionally, the blockchain industry’s primary fear has revolved around “Q-Day” — the speculative day when quantum computers can effectively dismantle public-key cryptography. Now, Drake posits that AI-driven mathematics could pose that risk sooner, potentially within “months not years” in the worst-case scenario.
“We should now prepare for the scenario in which ECDSA may fail even before Q-Day,” he noted, defining a successful breach as the recovery of a private key in roughly a week using readily available hardware, like a robust GPU cluster.
The catalyst for his alarm came from OpenAI’s release on October 6, which included 722 mathematical papers generated by an undisclosed internal model. These findings, shared on GitHub and spanning topics such as algebra, theoretical computer science, and mathematical logic, represent what Drake has termed “mathematical superintelligence.” He also pointed to a previous OpenAI test, where 10,000 autonomous AI agents worked collaboratively to solve the Navier-Stokes equation — a notoriously difficult problem in mathematics and physics — in just 88 hours.
“Recent days have challenged our understanding of human mathematical intuition,” Drake remarked. “Long-standing, accepted hypotheses have been overturned.”
His primary technical concern revolves around the intrinsic properties of elliptic curves. Unlike hash functions specifically designed to limit exploitable algebraic features, elliptic curves possess a “rich structure,” according to Drake, making them susceptible to intricate attacks. He highlighted that algorithms intended for quantum systems have, at times, led to new classical computing methodologies, suggesting that non-quantum strategies for breaching ECDSA could arise.
Drake’s suggested countermeasure is relatively straightforward in theory but complex in execution. When a wallet address signs a transaction, its public key becomes accessible on-chain. If attackers can deduce the private key from the public key, any assets still in that address become vulnerable. Conversely, new addresses that have never executed a transaction keep their public keys hidden behind cryptographic hashes.
He urged that significant and sophisticated holders should initiate this migration first, followed by others with smaller balances. However, he emphasized the importance of avoiding panic migration: “A hurried transition could create more problems than it solves,” he warned, noting that flawed transfers might lead to errors, compatibility challenges, and financial losses.
Drake specifically identified major custodians and exchanges — including Binance, Bitbank, Robinhood, Bitfinex, and Tether — as leaders essential to strengthening their cold storage protocols. He also referenced “Satoshi’s shield”: around 20,000 exposed addresses linked to Bitcoin’s pseudonymous founder, each holding approximately 50 BTC. He indicated that these wallets would likely become primary targets in any potential attack, providing a buffer for smaller holders.
Ethereum co-founder Vitalik Buterin responded to Drake’s caution with a tempered agreement. While endorsing the necessity to regard AI-enhanced cryptographic threats seriously, Buterin advised against immediate wallet migrations out of fear. “I don’t encourage anyone to rush and relocate their assets to new wallets now,” he stated.
Buterin extended the discussion further than Drake by addressing concerns beyond elliptic curves, highlighting lattice-based cryptography as a notable candidate for post-quantum security. He remarked that there exists a “good chance” that lattice-based systems could suffer “serious hits” within the next two years due to AI-enhanced mathematical developments. This raises doubts about the commonly held belief that even if elliptic curves fail, hashes and lattices will emerge unscathed.
The Ethereum Foundation has already started to formalize its strategy in response to these cryptographic threats. Earlier this year, the organization formed a dedicated post-quantum security group, and its draft roadmap is steering towards hash-based cryptography along with formal verification. Drake expressed his intent to advocate for rapid advancement along that timeline.
The broader cryptocurrency landscape has demonstrated related indicators. In July, Anthropic reported that its Claude Mythos Preview model enhanced the most effective attack on HAWK, a post-quantum signature candidate submitted to NIST, diminishing its effective key strength by 50% within 60 hours. Though HAWK is not yet in live systems, this result illustrates how AI can significantly affect cryptanalysis.
Europol also chimed in recently with a report indicating that blockchains themselves are not vulnerable to quantum computing hacks due to the resilience of their hash functions. The agency designated wallets as “the primary point of exposure,” suggesting proactive migration as the sole feasible solution — a conclusion that closely aligns with Drake’s recommendation.
The Europol report stated that nearly 6.04 million BTC — approximately 30.2% of the total supply — had public keys that were exposed as of May 2026.
Drake’s warning does not imply any immediate compromise regarding Bitcoin or Ethereum’s security. Current research does not indicate that ECDSA has been breached, and OpenAI’s mathematical release did not detail practical attacks on crypto systems. The researcher himself framed the proposal as precautionary rather than an immediate response.
Nevertheless, this incident signifies a pivotal moment in how the industry perceives cryptographic risks. The threat model is shifting from a singular focus on quantum technologies to encompass a broader spectrum, including algorithmic advances driven by AI. For industry participants, the pressing question is tactical: how to mitigate private key exposure via address rotation without introducing operational errors that could trigger losses.
Future indicators to monitor will include whether major wallet providers, custody firms, and exchanges adopt address-rotation protocols into their best practices — and if the Ethereum Foundation accelerates its roadmap for hash-based cryptography following Drake’s call for heightened defensive measures.