Claude reveals ECDSA curve origins
Anthropic says its model traced NIST curve seeds to a timestamp-seeded random-number generator.
TL;DR
- Anthropic says Claude identified the origins of five NIST elliptic-curve seeds.
- The company says the seeds came from the C library's lrand48 generator initialized with a Unix timestamp.
- Anthropic says the finding explains provenance questions but does not show a cryptographic break.
Anthropic reports that an internal research model identified how seeds for NIST curves P-192, P-224, P-256, P-384 and P-521 were generated. The company says the model reached the result in under half an hour. [1]
According to Anthropic, the seeds were outputs from the C library lrand48 pseudo-random number generator, initialized with a Unix timestamp. The post describes sequential 20-byte candidates and says P-256 took 1,842 iterations to match the published seed. [1]
Anthropic frames the result as resolving a long-running provenance question about parameters originally published by the NSA without an explanation. The finding concerns how the constants were generated, rather than a newly demonstrated weakness in the curves. [1]
The company explicitly says this is not a cryptographic break or a new attack and that no change in how the curves are used follows from the result. Independent cryptographic review of the reconstruction remains relevant to how broadly the explanation is accepted. [1] [2]
Why it matters
The story is a useful case study in AI-assisted cryptanalysis, while the source itself says the result does not change ECDSA security assessments.
Editor's note
Headline and technical claims checked against Anthropic’s post; Reddit is linked as discussion, not corroboration.