Claude helps trace the origins of cryptographic curve seeds

Anthropic researchers say Claude helped them trace the origins of seed values used in several long-standing cryptographic standards. Their finding points to a mundane explanation for values that had fuelled questions about whether the curves might contain a deliberate backdoor.

Anthropic Watch analysis

What happened

The researchers used Claude to investigate the origins of seed values in elliptic-curve standards published by NIST and ANSI. Anthropic says they traced the values to the lrand48 pseudorandom number generator, initialised with Unix timestamps, in 1990s C libraries.

The work offers an explanation for how the seed values were generated. Anthropic says that explanation increases confidence that the curves were selected without intentional backdoors. It does not, by itself, establish that every implementation using those standards is secure.

Why it matters

Cryptographic standards depend on confidence in choices that can be difficult to reconstruct decades later. A clearer account of how these values were produced helps address a persistent question about the provenance of widely used elliptic curves.

It is also a specific example of AI being used to investigate a technical mystery: Claude helped test hypotheses about historical software and connect the evidence to a plausible origin. That is more concrete than a general claim that AI can help with research, though the source does not provide enough detail here to assess how much of the investigation the model performed or how its conclusions were checked.

Our read

This is an intriguing use of Claude as a research assistant, and the finding gives cryptographers a useful explanation to examine. The key distinction is that tracing the seeds’ origin supports confidence in the selection process; it is not a blanket security certification for the curves or systems built on them.

What to watch

• Whether the researchers publish technical detail on their evidence and validation process.
• How cryptographers assess the proposed origin of the seed values.
• Whether this approach helps resolve other open questions about the history of cryptographic standards.

Discussion spark: How much should a plausible historical explanation increase confidence in a cryptographic standard, without a full account of how the finding was independently checked?

Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.