Anthropic's Claude Solves 25-Year Elliptic Curve Seed Mystery, Bitcoin Avoids NIST Curve

Huo Xing Cai Jing reports that on October 17, Anthropic disclosed that its internal model, Claude, cracked the 25-year-old mystery surrounding the seed generation of NIST standard elliptic curves in less than half an hour. This discovery is closely related to the foundational security design of cryptocurrencies. ECDSA is a widely used cryptographic standard for digital signatures, underpinning wallet signing and authentication for a vast number of crypto assets. Around 2000, certain parameters of the NIST standard elliptic curves were provided by the U.S. National Security Agency (NSA), but the seed generation process remained opaque for years, fueling speculation about potential backdoors. Through reverse analysis, Claude determined that the relevant seeds were most likely generated using the C language library’s pseudo-random number generator lrand48 combined with Unix timestamps—not by a complex, deliberately designed secret algorithm. This provides a new explanation for the long-unresolved origin of these parameters. Bitcoin, however, took a different path due to security concerns: Satoshi Nakamoto chose the secp256k1 curve instead of the commonly used NIST standard curves. This choice is often discussed in the context of a lack of transparency in standard-setting processes, but it cannot be simplistically assumed that avoiding an NSA backdoor was its sole motivation. It is important to emphasize that Claude cracked the method of generating certain standard curve parameters—not ECDSA itself—and did not uncover any new mathematical vulnerabilities in the curves. Reconstructing the parameter generation process helps assess backdoor suspicions but does not definitively prove that all related implementations are risk-free. This test also illustrates another dimension of AI’s role in the crypto industry: it not only has the potential to reshape cryptographic offense and defense dynamics but can also be used to audit historical cryptographic designs and validate security assumptions, offering new tools for security research in crypto infrastructure.