The cryptographic security of the Bitcoin and Ethereum networks is facing a new theoretical challenge, driven by the rise of artificial intelligence. An international collaboration of over a hundred researchers, supported by autonomous agents, has succeeded in reducing the computational resources by 86.1% required to execute a crucial stage of a future quantum attack. This work focused on optimizing algorithms capable, in theory, of deriving a private key from a public key by exploiting vulnerabilities in current signature systems such as secp256k1.
The research competition known as ECDSA.Fail helped drop the complexity score from 10.75 billion to approximately 1.26 billion, even surpassing previous estimates published by Google Quantum AI. Concretely, the best configurations identified now require 1,151 logical qubits to complete nearly 1.3 million operations. Although these figures demonstrate a notable improvement in algorithmic efficiency, it is imperative to emphasize that no private keys have been compromised and that no current computer hardware is capable of executing such a threat against digital assets.
The major implication of this study lies in the transformative role of artificial intelligence in fundamental research. By automating coding, testing, and the exploration of micro-optimizations, AI agents have significantly accelerated the pace of discovery, allowing human researchers to focus on strategic direction. This efficiency gain demonstrates that quantum risk depends not only on the evolution of hardware—physical quantum computers—but also on the growing sophistication of the software that could one day leverage those capabilities.
In light of these prospects, the crypto ecosystem is intensifying its preparations. While the NIST plans a transition to post-quantum security standards within the next decade, the private sector is not standing still. Consortia bringing together leading financial players and blockchain infrastructures are investing millions of dollars to anticipate the migration of funds toward reinforced protections. The stakes are now significant: as theoretical research moves forward in leaps and bounds, the technical and logistical challenge of effectively updating blockchain networks could span several years.