The Ethereum Foundation has unveiled better.codes, a collaborative research platform that applies artificial intelligence to strengthening the security of hash-based SNARKs. The initiative forms part of Ethereum's broader effort to prepare its cryptographic infrastructure for the potential risks posed by future quantum computers.

The platform centers on koalaIRS12, a Reed-Solomon proximity problem that plays an important role in modern cryptographic proof systems. Through better.codes, researchers can deploy AI agents to improve the soundness bounds of koalaIRS12 proofs, with results then subjected to verification through the Lean kernel. Lean, an open-source theorem prover and programming language originally developed at Microsoft Research, checks every proof step against a small trusted kernel, so accepted results carry machine-verified guarantees rather than relying solely on manual peer review.

The initiative aims to establish machine-verified security guarantees for koalaIRS12 while working toward a 128-bit security target that could strengthen Ethereum's defenses against future quantum threats. The project has been developed together with Yukon and zkSecurity and is structured around incremental contributions. Accepted solutions are credited both to the researchers or solvers and to the AI models involved in producing them — an approach intended to encourage transparent collaboration while creating a verifiable record of progress in cryptographic research.

Ethereum Prepares for Quantum Computing Risks

Ethereum's move toward post-quantum cryptography is driven by concerns over the long-term security of the cryptographic systems currently used across the blockchain. Elliptic-curve-based technologies underpin areas including transaction signatures, consensus infrastructure and zero-knowledge proofs.

A sufficiently capable quantum computer could potentially undermine some of these cryptographic primitives through Shor's algorithm, developed by mathematician Peter Shor in 1994, which targets the factoring and discrete-logarithm problems that underlie much of today's public-key cryptography. Although such machines are not currently capable of breaking Ethereum's widely used cryptography at practical scale — and major hardware developers such as IBM and Google have published roadmaps that target fault-tolerant quantum computers only toward the end of the decade — the potential future threat has encouraged blockchain developers to investigate alternatives ahead of time.

Ethereum has therefore been examining cryptographic approaches that rely on hash functions rather than elliptic-curve discrete logarithms. Hash-based SNARKs and STARKs are among the technologies being considered as part of this transition. The direction parallels standardization work beyond blockchain: in 2024, the U.S. National Institute of Standards and Technology finalized its first post-quantum cryptography standards, including SLH-DSA, a hash-based digital signature scheme, bringing hash-based cryptography into government-approved production use. The better.codes initiative addresses one of the challenges associated with that effort by seeking stronger and formally verified security guarantees for Reed-Solomon proximity codes — codes that underpin several cryptographic proof systems and are important to the development of scalable zero-knowledge technologies.

128-Bit Security Target

The better.codes challenge is focused on reaching a 128-bit security benchmark for koalaIRS12, a target consistent with Ethereum's longer-term objective of developing cryptographic systems capable of remaining secure in a post-quantum environment. The Ethereum Foundation announced the initiative on X:

Learn more in the announcement ↓

— Ethereum Foundation (@ethereumfndn) August 20, 2026

Full post: https://x.com/ethereumfndn/status/2090485717273043227?ref_src=twsrc%5Etfw

The work follows the Proximity Prize initiative, which was launched earlier in 2026 with a $1 million research fund dedicated to unresolved problems involving list decoding and correlated agreement. By expressing these cryptographic problems in Lean 4, the Ethereum Foundation is seeking to make research results easier to verify mechanically and potentially accelerate progress toward more reliable cryptographic constructions. The pairing of machine learning with formal proof checking also reflects a broader research trend — Google DeepMind's AlphaProof, which achieved silver-medal-standard performance at the 2024 International Mathematical Olympiad by producing Lean-verified proofs, illustrated how AI systems can contribute to formally checked mathematics.

Stronger verification of these cryptographic assumptions could benefit developers building rollups, zero-knowledge virtual machines and other scaling systems that increasingly depend on SNARK-based technology.

Long-Term Security Over Immediate Market Impact

The initiative is not expected to have a direct short-term effect on the price of ETH. Its significance instead lies in Ethereum's long-term protocol security and in preparations for technological risks that could emerge as quantum computing develops.

Recent efforts across the Ethereum ecosystem have likewise placed greater emphasis on post-quantum security, including proposals involving recursive STARK proofs and the establishment of a dedicated security team focused on quantum-related threats. For developers and infrastructure providers, these developments indicate that cryptographic resilience is becoming an increasingly important consideration as zero-knowledge technologies expand across the blockchain industry.

The better.codes platform is expected to evolve beyond its initial challenge, with the Ethereum Foundation planning additional research problems under program rules covering eligibility, evaluation and awards. Researchers can participate by signing in through GitHub, accessing the relevant repository and contributing solutions to the koalaIRS12 challenge.

As per reports, the initiative represents an effort to move post-quantum research beyond theoretical discussions and toward independently verifiable implementation. By combining AI-assisted research with formal Lean verification, better.codes could provide Ethereum with a more structured pathway toward cryptographic systems designed to withstand future quantum-era threats.

Source: CoinTrust