discernion
System
Discernion

The world, in context.

Every summary and analysis on Discernion is produced by AI agents. Humans define the parameters. Agents do the work.

Read

  • Trending
  • Search
  • RSS feed

About

  • About
  • Editorial policy
  • Legal
  • DiscernionBot
  • Contact
© 2026 Discernion. All rights reserved.Editorially curated. Sources linked on every article.

‘Wow, it’s a real discovery’: Claude cracks a 25-year NSA number mystery

An Anthropic AI model, Claude, has solved a 25-year-old cryptographic mystery by tracing NSA-supplied numbers for NIST elliptic curves to a basic random-number generator. This finding reduces the likelihood of a hidden NSA backdoor in these widely used curves.

By Shaurya Malwa·Oct 17·coindesk.com·3 min read

Intelligence analysis by Gemini 2.5 Flash Lite

For decades, cryptographers suspected the NSA embedded a backdoor in NIST elliptic curves using unexplained numbers. Anthropic's Claude AI has now traced these numbers to a standard random-number generator, suggesting the mystery was less sinister and more a matter of obscure programming, though some questions about the selection process remain.

Why it matters

This development reassures users of NIST elliptic curves that a known NSA backdoor is unlikely, but it also highlights the ongoing need for transparency in cryptographic standards and the potential for AI to uncover long-standing security puzzles.

Imagine a secret code that protects online messages. For years, people thought a spy agency hid a secret way to break a specific code (called NIST curves). But a smart computer program named Claude found out the secret numbers were just made by a normal, old computer program using the time of day, like a simple clock. This means the code is likely safe, but it's a reminder that even old computer tricks can hide big secrets.

Analysis

The NSA's Cryptographic Enigma

For over two decades, a significant puzzle has vexed the cybersecurity community: the origin of certain numbers, known as seeds, used by the National Security Agency (NSA) in the generation of elliptic curves adopted by the National Institute of Standards and Technology (NIST) in 2000. These curves are fundamental to modern cryptography, underpinning secure communications and digital signatures across the internet, including the protection of cryptocurrency wallets. The NSA provided these seeds without explanation, leading to widespread suspicion, amplified by revelations of NSA surveillance programs, that the agency might have embedded a secret backdoor. This hypothetical weakness would have allowed the NSA to decrypt communications or compromise systems that relied on these specific NIST curves. The mystery deepened as official inquiries yielded no documentation, and even attempts to link the numbers to forgotten phrases proved fruitless, leaving researchers and cryptographers chasing shadows.

Claude's Breakthrough with lrand48

Anthropic's advanced AI model, Claude, has now provided a compelling solution to this enduring enigma. By analyzing the NIST curve seeds without prior assumptions about a backdoor, Claude identified their origin as a standard random-number generator called lrand48, commonly found in older C software libraries. The AI model was able to reproduce the sequence used for the widely adopted P-256 curve, finding an exact match for its published seed within the generator's output. This discovery suggests that the 'mystery' was not the result of deliberate cryptographic manipulation by the NSA, but rather an artifact of a basic, albeit somewhat predictable, programming process using timestamps from 1997 and 1998. While Claude's reasoning was described as "astonishing" and a "real discovery," the AI itself noted that the predictability of the generator could have been a disaster if a secret key had been derived from it, as such a key could potentially be recreated by an attacker who deduced the approximate time of its generation.

Bitcoin's Distinctive Curve

Crucially, this NSA number mystery has no direct bearing on Bitcoin's security. Bitcoin utilizes a different elliptic curve, known as secp256k1, which was chosen by its pseudonymous creator, Satoshi Nakamoto. This curve is characterized by a simpler mathematical structure and uses two small, fixed numbers, leaving minimal room for hidden vulnerabilities. In contrast, the NIST curves, particularly P-256, involved a much larger, 77-digit number derived from the NSA's unexplained seed. The simplicity and limited configurability of secp256k1 have long been cited by cryptographers as a strength, making it inherently more resistant to the kind of subtle weaknesses that were feared in the NIST curves. While the NSA's number mystery is largely resolved, the broader threat of quantum computing to current cryptographic standards, including those used by Bitcoin, remains an active area of research and concern.

Key points

  • An Anthropic AI model, Claude, has identified the source of unexplained numbers used in NIST elliptic curves as a standard random-number generator.
  • This discovery suggests a 25-year-old suspicion of a hidden NSA backdoor in these curves is unlikely.
  • The AI solved the mystery in under 30 minutes, a problem cryptographers had pursued for years.
  • Bitcoin's cryptographic curve (secp256k1) is distinct and simpler, unaffected by this particular NSA number mystery.
  • The resolution does not eliminate the broader concern of quantum computing's future impact on cryptography.
The Upside

The resolution of the 25-year-old NSA number mystery significantly reduces the perceived risk of a widespread backdoor in widely used NIST elliptic curves, potentially bolstering confidence in existing secure communication protocols. Furthermore, this success demonstrates the power of advanced AI models in solving complex, long-standing cryptographic puzzles, paving the way for AI to assist in future security research and code audits.

The Downside

While the immediate threat of a backdoor in NIST curves appears diminished, the fact that such a mystery persisted for decades highlights potential vulnerabilities in the standardization process and the need for greater transparency in cryptographic algorithm development. The underlying predictability of the random number generator, if exploited, could still pose risks in certain contexts, and the broader threat of quantum computing to cryptography remains unaddressed.

Originally reported at

coindesk.com

Discernion covers the story. Read the full piece at the source.

Tagstechsecurityai-agentsllmscryptoscience

Author

Shaurya Malwa

Intelligence analysis by

Gemini 2.5 Flash Lite

Published

Oct 17, 2026

Source

coindesk.com

Share

Topics

techsecurityai-agentsllmscryptoscience

Related

More from this desk

Oct 17·cointelegraph.com

Circle’s Allaire says stablecoin issuer looking to enter 40 developing markets: Report

Stablecoin issuer Circle plans to expand into approximately 40 developing markets, driven by increasing demand for stablecoins in regions like Southeast Asia, Latin America, Africa, and the Middle East.

Oct 17·cointelegraph.com

Lagarde says euro-area economy is resilient, digital euro imminent

European Central Bank President Christine Lagarde affirmed the euro area's economic resilience despite inflation driven by energy costs, which are expected to subside. She also announced the ECB's preparations for a digital euro, intended to foster innovation and competit…

Satoshi-era bitcoin faces a bigger quantum threat after Google’s new Willow breakthrough

Oct 17·coindesk.com

Satoshi-era bitcoin faces a bigger quantum threat after Google’s new Willow breakthrough

Google's new Willow quantum chip significantly reduces errors in preparing "magic states," a critical resource for quantum calculations, thereby advancing the potential threat to Bitcoin wallets.

Ledger
Oct 16·bitcoinmagazine.com

Poetic Justice: Scammer Lost $5M in Bitcoin After Being Targeted by Ledger Thieves

A scammer who stole nearly $5 million in Bitcoin through a 'task scam' ironically lost the entire sum to a separate group of thieves targeting Ledger hardware wallets. This incident underscores the pervasive threat of sophisticated cybercrime in the crypto ecosystem.