Finding the source of US standard cryptographic curve seeds in time-based random numbers (anthropic.com)

Writing language: KoreanRead in the original language

- ECDSA is a widely used cryptographic standard for things like digital signatures. The seeds of the curves NIST adopted in 2000 were published without any explanation of their origin, and they have been doubted for decades. - Anthropic said that an internal research model found the origin of the P-192, P-224, P-256, P-384, and P-521 seeds in about 30 minutes without hints. This is the company's own account, and the source does not include any outside verification. - According to the company, the seeds are not hash values. They are output from the C library random number generator lrand48, seeded with Unix time. The source publishes code that reproduces the same calculation. - The source states that this is not a cryptographic vulnerability or a new attack. However, it notes that the origin of some base points (the starting coordinates of the curves) and the seed of the binary curve B-163 remain unexplained.

Found on

Hacker News ↗ / 808 votes / 306 comments (checked 2026-10-17 15:39 KST)

Have you run into something like this? Tell us in a line or two what you did. You do not need to know the product.

1 comment

Editorial opinionstartupxo 25m ago

O: This is useful for a security engineer who uses NIST curves and wants to run the reproduction code from the original article and compare the seed output directly while reviewing internal cryptography configuration documents. X: This result was produced and interpreted by Anthropic. The original article also says it could not explain how the skipped candidate curves were selected or where the B-163 seed came from, so that part should not be treated as settled until it is verified.

Writing language: Korean

Related reads

1 comment