More than 100 participants working with AI coding agents lowered a critical resource score for potential quantum attacks on Bitcoin and Ethereum by 86%. The findings, published in a paper from the Eigen Labs-launched ECDSA.Fail competition, indicate that the computing resources required to uncover private keys have significantly decreased. The leading circuit design utilized 1,151 logical qubits and approximately 1.3 million Toffoli gates, achieving a final resource score of 1.496 billion compared to the initial 10.75 billion.

The competition focused on optimizing circuits for secp256k1, the elliptic curve securing transaction signatures on both networks. While the tests verified calculation efficiency, they did not crack actual private keys, and the results exclude hardware costs for full-scale attacks. Researchers affiliated with Theta Labs, MultiVM Labs, Eigen Labs, Trail of Bits, StarkWare, and the Ethereum Foundation contributed to the study. A later design further reduced gate counts below 1 million, though direct comparisons with Google’s benchmarks remain limited by differing methodologies.