Key highlights:
- Charles Hoskinson challenged Vitalik Buterin’s warning that AI could weaken lattice-based cryptography, arguing that the threat remains speculative
- The Cardano founder questioned Buterin’s proposal to increase lattice-cryptography security parameters tenfold, demanding a concrete attack model and defensible cost estimates
- Hoskinson argued that hash-based cryptography also faces potential vulnerabilities and warned against abandoning established post-quantum security systems without evidence of a breakthrough
Cardano founder Charles Hoskinson has launched a detailed technical rebuttal against Ethereum co-founder Vitalik Buterin’s warning that artificial intelligence could weaken lattice-based cryptography within two years. Hoskinson argued that Buterin’s proposed shift toward hash-based cryptography rests on speculative attack scenarios and an unjustified recommendation to increase lattice security parameters tenfold.
Hoskinson challenges Vitalik Buterin’s cryptography argument
Hoskinson disputed Buterin’s comparison between potential AI-driven attacks on lattice cryptography and the mathematical breakthrough that produced the General Number Field Sieve, an algorithm used to factor large integers.
Buterin warned on October 8 that AI could accelerate mathematical research enough to undermine lattice-based cryptographic systems. He recommended favoring hash-based alternatives where possible and adopting more conservative parameters for lattice-based schemes.
Hoskinson argued that the factoring analogy fails to identify a comparable mathematical breakthrough that could weaken lattice cryptography. He pointed to decades of research into lattice reduction, including the Lenstra–Lenstra–Lovász algorithm, blockwise lattice reduction and increasingly sophisticated sieving techniques.
According to Hoskinson, researchers have incorporated the costs of these attacks into the security parameters of standardized lattice-based systems. He argued that any proposal to increase key sizes tenfold should identify a concrete attack and explain its computational cost.
Cardano founder questions the case for hash-based security
Hoskinson also challenged the assumption that hash-based cryptography offers a safer alternative simply because its underlying functions appear to have less exploitable mathematical structure.
He pointed to attacks that compromised MD5 and SHA-1, demonstrating that hash functions can develop practical weaknesses when researchers discover exploitable properties in their designs. Hoskinson also questioned the security assumptions behind hash-based proof systems and Poseidon, a cryptographic hash function designed for efficient use in zero-knowledge proofs and other cryptographic applications.
Source: pqsignatures.org
Hoskinson argued that AI-driven mathematical advances could affect hash-based systems as well as lattice-based ones. He maintained that security recommendations should account for the specific algorithms, attack costs, and parameters involved rather than assume one cryptographic family will remain safer than another.
The Cardano founder further argued that lattice cryptography has undergone extensive scrutiny and that researchers have already investigated structural weaknesses in specialized lattice constructions. He distinguished those findings from the standardized module-lattice schemes used in ML-KEM and ML-DSA, which he said have not been broken by the cited attacks.
“Vitalik is now trying to convince everyone that lattices are bad because he is bag-holding too much hash-based crypto research can’t back out,” wrote Hoskinson.
Debate raises questions about post-quantum security
The disagreement follows Buterin’s warning that AI-assisted mathematical discoveries could threaten cryptographic systems sooner than the industry expects. In his October 7 post, the Ethereum co-founder identified lattice-based digital signatures, fully homomorphic encryption and other systems as areas requiring greater caution.
Buterin also advised against rushing to migrate cryptocurrency holdings into new wallets, warning that mistakes during hurried transfers could cause losses. He argued that the industry should prepare for potential AI-driven attacks while reducing exposure to cryptographic systems that could become vulnerable.
Hoskinson took issue with the practical implications of Buterin’s recommendations, arguing that abandoning established lattice-based systems without a demonstrated attack could undermine confidence in post-quantum security research. He also warned that unnecessarily increasing security parameters could impose performance costs on systems that rely on cryptographic operations.