Crypto's Quantum Defense Is Being Built in the Open

What it is

Eigen Labs' Yukon (launched August) hosts open challenges. A problem is published, participants and AI agents submit solutions, each is tested automatically, and accepted improvements become the baseline for the next attempt. The article is Bankless's account of the quantum-related challenges and the non-quantum ones that show the model works.

Quantum-related challenges, with details the summary lacks

• ECDSA.fail: Started from Google's published resource-estimate verifier. Google withheld its optimized circuits. Participants design circuits for a key step of a quantum attack on Bitcoin and Ethereum cryptography. The combined resource-cost benchmark is down roughly 88%.
• HashSmash: Launched October 5 with Zcash, Shielded Labs and Project Eleven. Accepted attacks so far hit only simplified SHA-256, not the full algorithm. More ambitious submissions are awaiting review.
• sig.golf: Built with the Ethereum Foundation. It optimizes hash-based signatures, aiming for smaller and cheaper-to-verify signatures with a mathematical security proof. The leading result nearly halves a combined signature-size and verification-cost measure.
• Quantum Safe Bitcoin (StarkWare): Targets the GPU cost of a method that protects Bitcoin transfers against quantum attacks without changing Bitcoin's rules. The first mainnet demonstration in August cost roughly $320. The estimate is now $59, an 82% drop, but the cheaper version has not been run on a live transaction.
• better.codes: Built with the Ethereum Foundation. It uses AI agents to strengthen the proofs behind post-quantum cryptography. Provable security is up from 64 to about 68 bits, against a 128-bit target.

Evidence the model works outside quantum

These are benchmark results, some deployed:

• Lighter.fast: Roughly 10x faster in benchmarks. The first optimizations shipped in August and cut the exchange's real proving time in half.
• MLX.fast: Qwen about 2x faster and Poolside's Laguna 2.6x faster on Apple hardware, now in Eigen's Darkbloom inference network.
• SNARK.fast: With Succinct and the Ethereum Foundation, a calculation used in post-quantum proofs is about 3.5x faster in early benchmarks. Succinct has merged it into its codebase.

What to do

• If you maintain hash-based or post-quantum signature code, or SNARK provers, the sig.golf, better.codes and SNARK.fast results are the ones worth checking against your own parameters.
• Yukon accepts new challenge proposals, and results for HashSmash are published on an ongoing basis. The article links to them but gives no URL.

Caveats

• All quantum-attack gains are theoretical. No quantum computer has demonstrated an attack at this scale.
• The "88%" figure is a combined benchmark relative to the challenge's own baseline, not a cut in real-world attack cost.
• The $59 Bitcoin figure is an estimate, not a demonstrated result.
• The 64-to-68-bit proof gain is far from the 128-bit target.
• The article is promotional toward Yukon and gives no independent verification of the numbers.