Cardano founder Charles Hoskinson publicly criticized Ethereum co-founder Vitalik Buterin on Oct. 9, arguing that Buterin's speculative warnings regarding artificial intelligence undermining lattice-based cryptography could delay the implementation of quantum-resistant internet defenses.
Key Takeaways:
- Hoskinson challenged Buterin's pivot away from lattice-based designs toward hash-based signatures for Ethereum's long-term roadmap.
- The US National Institute of Standards and Technology (NIST) standardized ML-KEM and ML-DSA in 2024 after four decades of lattice research.
- Hoskinson dismissed Buterin's suggestion to multiply key sizes by ten as "numerology," arguing that hash functions like MD5, SHA-1, Poseidon, and Poseidon2 also carry structural risks.
Dispute Over Lattice-Based Cryptography
In an Oct. 9 post, Charles Hoskinson rejected Vitalik Buterin's assertion that AI-assisted mathematical discoveries could rapidly break lattice-based cryptography. Buterin previously argued that advances similar to the General Number Field Sieve (GNFS) in integer factorization could be accelerated by AI, prompting Ethereum to prioritize hash-based signatures and proofs in its lean roadmap. Hoskinson countered that lattice-based systems rely on more than 40 years of cryptanalytic research, including progress in sieving algorithms and lattice reduction.
He noted that NIST standardized ML-KEM for key encapsulation and ML-DSA for digital signatures in 2024 with security parameters specifically calibrated against known attack vectors. As developers work on protocol security across decentralized networks and layer-2 infrastructure, Hoskinson stressed that abandoning established lattice approaches could leave systems vulnerable to impending quantum threats.
Hash-Based Signatures and Structural Vulnerabilities
Hoskinson also disputed Buterin's suggestion of multiplying key sizes by ten to counter AI breakthroughs, calling the idea "numerology" and maintaining that algorithm parameter adjustments must rely on measured attack efficiencies. Furthermore, Hoskinson questioned whether Ethereum's preferred hash-based strategy provides superior safety, citing historical weaknesses in MD5 and SHA-1 to demonstrate that hash functions can contain exploitable structures.
He extended his criticism to Poseidon and Poseidon2, hash functions designed for zero-knowledge proofs that have received research funding from the Ethereum Foundation to evaluate resistance against algebraic attacks such as Gröbner bases. Hoskinson argued that hash-based constructions present similar targets for AI discoveries, warning that abandoning lattice research will severely limit developer options for encryption and privacy applications.
Why It Matters
The disagreement highlights a fundamental strategic split in how major blockchain networks plan to transition away from vulnerable elliptic-curve signatures before quantum computing matures. While Ethereum prioritizes a lean, hash-based roadmap, Hoskinson contends that discarding lattice-based designs risks restricting essential cryptographic tools like key encapsulation across broader internet infrastructure. As developers prepare for post-quantum standards, the debate underscores the challenge of balancing long-tested mathematical frameworks against emerging AI-driven cryptanalytic capabilities.
