Hoskinson challenges Buterin’s doubts about lattice cryptography

Cardano founder Charles Hoskinson challenged Ethereum co-founder Vitalik Buterin’s concern that artificial intelligence could help researchers find new attacks against lattice-based cryptography.

In an Oct. 9 post, Hoskinson argued that decades of research had tested the known weaknesses of lattice systems. He warned that discouraging developers from using them could slow the deployment of defenses against quantum attacks.

Buterin has argued that Ethereum’s long-term “lean” roadmap should favor hash-based signatures and proofs instead of lattice-based designs. His concern draws on the history of cryptography, including the development of the general number field sieve, which improved attacks against integer-factorization systems such as RSA.

Buterin suggested that AI could produce similar mathematical advances in a shorter period and expose weaknesses in lattice problems that researchers have not found. Hoskinson rejected that comparison.

“The case against lattices is the GNFS story, a.k.a. a hunch about ‘structure,’ and a multiplier pulled out of thin air,” Hoskinson wrote.

Hoskinson noted that no broad technique comparable to the general number field sieve has been demonstrated against the lattice problems used in current post-quantum standards. He pointed to more than 40 years of research into lattice reduction and sieving methods, which has improved attacks without producing a general method for breaking properly configured systems.

The U.S. National Institute of Standards and Technology standardized ML-KEM for key encapsulation and ML-DSA for digital signatures in 2024. Both use lattice-based mathematics. Hoskinson said their security parameters account for attacks identified through that research.

Hoskinson also criticized Buterin’s suggestion that multiplying key sizes by 10 could help protect public-key encryption from AI-assisted mathematical advances. He described the idea as “numerology” and argued that security parameters should change in response to measurable improvements in attack methods.

A moderate improvement in attack efficiency could require larger parameters, Hoskinson wrote. A fundamental mathematical breakthrough could require replacing an algorithm.

He also questioned whether hash-based cryptography would be less exposed to unexpected discoveries. Hoskinson cited weaknesses found in MD5 and SHA-1, while acknowledging that those failures do not show flaws in newer systems such as SHA-256. He also raised questions about Poseidon and Poseidon2, hash functions designed for efficient use in zero-knowledge proofs.

The disagreement concerns alternatives to the elliptic-curve signatures used by Bitcoin and Ethereum. Hash-based signatures can protect transaction authorization from quantum attacks. Lattice-based systems can support key encapsulation for encryption, privacy tools, secure communications and other cryptographic applications.