Anthropic Traces NIST P-256 Seed to a 1997 Random Generator

A decades-old mystery surrounding the origins of a widely used internet security standard has been resolved by researchers at Anthropic, who used the Claude AI model to trace the NIST P-256 elliptic curve seed to a conventional pseudorandom number generator run on August 16, 1997. The findings were announced on October 16, 2026, and offer evidence against longstanding suspicions that the curve's parameters concealed a deliberate security weakness.

According to Anthropic's findings, the seed, C49D360886E704936A6678E1139D26B7819F7E90, was generated using lrand48/drand48, a pseudorandom number generator commonly available on Unix systems. The process began with Unix timestamp 871696354, corresponding to August 16, 1997, at 01:52:34 UTC.

The researchers identified the seed as candidate number 1842 produced during that generator run. Their analysis suggests that the parameter originated from an ordinary software process rather than a specially selected value designed to conceal a cryptographic backdoor.

The investigation also identified timestamp patterns associated with other elliptic curves. According to the findings, the seeds for the prime curves P-192, P-224, P-384 and P-521 corresponded to dates between May 1997 and April 1998. Seeds for binary curves showed patterns consistent with dates in February 1984 or April 1998.

Anthropic's researchers reportedly recovered the generation method in less than 30 minutes by examining the behavior of pseudorandom number generators used during the 1990s. The team completed a challenge established by cryptographer Filippo Valsorda, which carried a $36,864 reward, and donated the proceeds to the Internet Security Research Group (ISRG).

The origins of the seed had attracted scrutiny because of the National Security Agency's involvement in cryptographic standardization. Jerry Solinas, a mathematician at the NSA, submitted the seeds to the ANSI X9.62 committee in fall 1997 while the committee was developing the Elliptic Curve Digital Signature Algorithm specification.

The National Institute of Standards and Technology subsequently standardized P-256 in 2000 through FIPS 186-2, building on the earlier ANSI X9.62 work. The lack of publicly documented explanations for the seed's origin fueled questions about whether its selection could have introduced an intentional weakness.

Those concerns intensified following the 2013 revelations about Dual_EC_DRBG, a pseudorandom number generator associated with controversy over potential cryptographic backdoors and the NSA's role in security standards. Earlier claims suggested that the curve seeds were SHA-1 hashes of humorous phrases, but researchers had been unable to reproduce that explanation.

Anthropic's analysis found no evidence of intentional cryptographic weaknesses in the seed-generation process. The researchers' explanation relies on the limited range of possible outputs associated with a conventional generator initialized at a specific timestamp, making deliberate selection of a seed with a hidden exploitable property less plausible.

However, the findings also highlight shortcomings in the original documentation. The seed appears to have been generated using an ordinary tool at an arbitrary moment, without a publicly recorded explanation of the procedure. Modern cryptographic practice favors transparent parameter-generation methods, including procedures designed to demonstrate that values were not selected to create hidden weaknesses.

The discovery does not change the P-256 curve or require modifications to existing implementations. The curve, also known as prime256v1, remains widely used in internet security. The investigation instead provides a documented explanation for a parameter whose origin had remained uncertain for decades.

Follow suaragarut.id Add to preferred sources on Google

• Artificial Intelligence
• Cybersecurity
• Cryptography