AIToday
Large Language ModelsAI Safety & AlignmentGIGAZINE AIPublished: Oct 19, 2026, 19:01 JST

Claude traces ECDSA seeds to 1990s lrand48 random numbers

Claude traces ECDSA seeds to 1990s lrand48 random numbers

3 Key Points

  1. What happened

    Anthropic said Claude, given the task in October 2026, found in under 30 minutes that NIST's five ECDSA curve seeds came from the 'lrand48' pseudo-random generator, seeded with 1997–1998 Unix timestamps.

  2. Why it matters

    Because the seeds trace to an ordinary random generator, the long-suspected chance of a deliberate backdoor is now considered lower than before — and Anthropic says the finding does not show ECDSA is vulnerable.

WHO IT HITSSecurity researchers and cryptography engineers who rely on NIST's elliptic-curve standards gain a documented origin for the seeds, while organizations using ECDSA certificates see no indicated need to change them based on this finding.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

FAQ
Does this mean ECDSA is broken or unsafe?
No. Anthropic says the finding does not demonstrate a vulnerability in ECDSA; it only identifies how the seed values were generated.
How did Claude reproduce a specific curve's seed?
For the P-256 curve, it generated random numbers from a starting value of August 16, 1997, 01:52:34 UTC and took the 1,842nd candidate when split into 20-byte chunks.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleClaude Opus 4.7 'Eko' plays alone on island, learns better