The finding

An AI model named Claude took less than 30 minutes to trace five unexplained seeds used in NIST elliptic curves to an ordinary random-number generator seeded with timestamps from 1997 and 1998. The result, described by one observer as a real discovery, addresses a question that had hung over modern cryptography for about 25 years: whether the numbers were selected to conceal a backdoor for the National Security Agency.

The seeds were part of the process used to define elliptic curves standardized by the National Institute of Standards and Technology. Those curves are embedded in web encryption, secure messaging, software signing, identity systems and many other security products. If the parameters had been deliberately chosen to hide a mathematical weakness, the consequences could have been severe. The new finding makes that scenario less likely, though it does not answer every question about how the standards were assembled.

Key facts

• Claude, an AI model, took under 30 minutes to trace five unexplained seeds for NIST elliptic curves to a basic random-number generator.
• The generator used timestamps from 1997 and 1998, according to the finding.
• The discovery makes a hidden NSA backdoor less likely.
• Users of NIST curves do not need to take action based on the result.
• Questions remain about how some elements of the curves were selected.
• Bitcoin was not affected because it uses secp256k1, not the NIST curves in question.
• Quantum computing remains a separate, longer-term risk for elliptic-curve cryptography, including Bitcoin.

Why five seeds mattered

Elliptic-curve cryptography relies on mathematical objects called curves, which are defined by a set of parameters. A user does not need to know every detail of those parameters to use the system, but cryptographers do. If a designer chooses parameters that look random but are built around a secret trapdoor, an attacker who knows the trapdoor may be able to recover private keys or forge signatures. This is why standards bodies try to generate parameters in a transparent, reproducible way. When a number cannot be explained, the obvious question is whether it was chosen for an ordinary reason or for a hidden one.

The five seeds were especially sensitive because they were associated with curves that NIST standardized with input from the NSA. For years, cryptographers could see the resulting numbers but could not fully explain their origin. That gap in the public record turned into a durable suspicion. It did not prove wrongdoing, but it kept the issue alive in security reviews, academic papers and engineering discussions.

A quarter-century of suspicion

The concern did not arise in a vacuum. During the 1990s, the U.S. government fought a series of public battles over encryption. Officials worried that strong cryptography would hinder law enforcement and intelligence work, while civil-liberties advocates and technologists warned that weak or compromised standards would harm everyone. The Clipper chip proposal, export controls and debates over key escrow all reinforced the idea that government agencies might try to shape security standards in ways the public could not easily verify.

Those suspicions deepened after the 2013 disclosures about surveillance programs and the Dual_EC_DRBG random-number generator. That generator was standardized by NIST and later shown to contain a structure that could allow an attacker with a secret value to predict its output. The episode became a cautionary tale about opaque constants and agency influence. Even though the NIST elliptic curves are a different part of the standards, the controversy made unexplained parameters look more dangerous than they might otherwise have seemed.

What the model did

According to the account, Claude approached the problem as a search over plausible generation methods. It tested whether the five seeds could have come from a simple random-number generator rather than from a deliberately engineered construction. The model then matched the values to timestamps from 1997 and 1998. That was enough to connect the seeds to an ordinary computational process and to reduce the likelihood that they were chosen to hide a backdoor.

The speed of the result is part of the story. A task that might have taken a human researcher considerable time, or might never have been attempted with the same breadth, was completed in under half an hour. The finding still needs to be understood in context, but it demonstrates how AI tools can assist with cryptographic forensics by generating and testing hypotheses at scale.

What changes and what does not

The immediate practical takeaway is reassuring: users of the NIST curves do not need to take action because of the finding. The result does not identify a flaw in the curves, and it does not mean that every deployment is automatically safe or unsafe. It simply makes one long-suspected explanation, a hidden NSA backdoor, less likely.

That does not mean the standards are now fully transparent. Questions remain about how some elements of the curves were selected. A discovered seed origin can explain one part of the process without explaining every choice. Cryptographers may still want a complete, reproducible account of the parameter-generation procedure, including the order of operations, the exact inputs and the reasons for any deviations from a purely random process. Trust in cryptography depends on verifiability, not just on a plausible story.

Bitcoin was never on the list

Bitcoin was not affected because it does not use the NIST curves in question. Its signatures rely on secp256k1, a different elliptic curve with a simpler structure and very little room to conceal a weakness. That design choice has long been part of Bitcoin's appeal to users who prefer to avoid standards shaped by intelligence agencies. The new finding therefore does not change Bitcoin's security model.

It is still important not to overstate the point. A simpler curve is not automatically immune to every flaw, and the security of Bitcoin depends on many other components, including key generation, software implementation and user behavior. But with respect to this particular backdoor suspicion, Bitcoin sits outside the affected set.

The quantum question remains

A separate risk has not gone anywhere: sufficiently powerful quantum computers could eventually break elliptic-curve cryptography, including the curve used by Bitcoin. The relevant algorithm is well known in theory, but the engineering challenge is enormous. No public quantum computer today is close to the scale needed to threaten deployed elliptic-curve systems. Still, the possibility has prompted research into post-quantum cryptography and long-term migration planning.

For Bitcoin, a future quantum threat would raise difficult governance and infrastructure questions. The network would need to move users toward quantum-resistant signatures while preserving compatibility and avoiding disruption. That is a different problem from the NSA number mystery, and it is not solved by showing that the NIST curve seeds came from an ordinary generator.

Remaining questions

The investigation leaves several questions for cryptographers and standards watchers. How exactly were the five seeds incorporated into the curve-generation process? Were there other unexplained inputs that deserve similar scrutiny? Can the result be independently reproduced with the same timestamps and generator? And what does the episode say about how standards should be documented so that future researchers do not have to rely on inference?

Those questions matter because cryptographic trust is cumulative. A single unexplained number does not automatically create a backdoor, but it can create doubt. A single explained number does not automatically remove all doubt, but it can narrow the space in which a flaw might hide. The finding therefore moves the conversation from suspicion toward verification. The next step is to see whether the method can be applied to other long-unexplained constants and whether the full selection history can be reconstructed in public.

Source:Coindesk News