Ethereum co-founder Vitalik Buterin recently issued a warning about the potential risks posed by AI-driven advancements in mathematics on existing crypto signatures and post-quantum defenses. In a tweet on Oct. 7, Buterin advised crypto holders to be cautious and reduce exposure where possible, but also urged against hastily moving funds to new wallets. He emphasized the importance of taking the risks associated with AI-accelerated math seriously.
The discussion was sparked by Ethereum researcher Justin Drake’s suggestion that the industry should start preparing for a “bunker mode” scenario by moving assets to fresh addresses with public keys that have never been exposed on-chain. Drake raised concerns about the potential for AI to accelerate mathematical attacks on the elliptic-curve cryptography used by Bitcoin and Ethereum.
The debate gained traction following the publication of new mathematical results by OpenAI on Oct. 6, generated by an internal frontier model. While these results did not indicate any attacks on blockchain cryptography, Buterin and Drake highlighted the possibility of AI-driven mathematical discoveries weakening the security margins of lattice-based cryptography, including ML-DSA.
Buterin expressed concerns about the underlying mathematical structures of lattice-based systems, suggesting that AI advancements could uncover vulnerabilities that have yet to be discovered. He drew parallels to the history of integer factorization, where advancements like the general number field sieve significantly reduced the work required to attack RSA, leading to the adoption of larger keys for cryptographic systems.
The potential compression of decades of mathematical progress into a shorter period by AI could result in breakthroughs that enhance attacks on lattice cryptography. This poses a significant challenge for an industry already preparing for the advent of quantum computers, as lattice-based systems are considered a leading technology to replace cryptography vulnerable to quantum machines.
In response to these concerns, Ethereum developers have been leaning towards cryptographic designs based on hash functions. Buterin noted that Ethereum’s roadmap has shifted towards hash-only systems over the past year, avoiding lattice-based signatures like ML-DSA and Falcon. Hash-based signature systems, such as WOTS and SPHINCS, are favored due to their reduced reliance on vulnerable mathematical structures.
While hash-based systems offer a more secure alternative, they do not address all cryptographic challenges. Systems requiring public-key encryption, such as secure messaging and anonymous communications, rely on mathematical structures that may need larger security parameters in the face of AI-driven cryptanalytic advancements.
In light of these developments, Buterin advised crypto holders to take precautions to minimize exposure, such as keeping funds in addresses that have never signed a transaction. He also suggested that multisig operators rotate signer keys after operations to limit the usefulness of exposed keys if ECDSA security weakens.
Overall, the crypto industry faces a complex challenge of balancing the need for heightened security measures against the risks of operational failures. As the pace of cryptographic risk accelerates, stakeholders must navigate these uncertainties while ensuring the integrity and security of their platforms and assets.
