J. Ayo Akinyele, the head of engineering at Ripple, has responded to a significant cryptographic discovery made by Anthropic's Claude AI that reportedly resolved a 25-year-old enigma about the origins of popular cryptographic standards
Akinyele commented that the findings were quite remarkable, noting that the solution was actually quite straightforward
"AI shedding light on the origins of the NIST EC seeds is a pretty remarkable result," Ayo Akinyele said on X, citing the revelation that the origin of the parameters involved the use of the lrand48 random number generator seeded with the Unix timestamp
The recent breakthrough relates to the origins of the widely debated elliptic curve parameters standardized by the National Institute of Standards and Technology (NIST), which have long been subject to speculation regarding potential backdoors inserted by the US National Security Agency (NSA)
He said the breakthrough offers reproducible evidence and greater confidence about the origins of such cryptographic standards
Akinyele went on to speculate about other mysteries in cryptography that could potentially be solved with the help of artificial intelligence
In an October 16 research report, Anthropic announced that their internal version of Claude discovered the origins of five popular NIST elliptic curve seeds within half an hour
They include P-192, P-224, P-256, P-384, and P-521, which are part of the standards for the Elliptic Curve Digital Signature Algorithm (ECDSA)
The origins of the original seed values had been a subject of curiosity among researchers for decades, in part due to the fact that the seeds were provided by the NSA with little detail about their provenance
However, Claude found that lrand48, which is a fairly simple pseudorandom number generator included in older C programming libraries, was used to generate the seeds
It is worth pointing out that this is not a successful attack on ECDSA. Nor does it mean there are any security concerns for cryptocurrencies that use elliptic curve cryptography
Rather, it makes the theory of intentional weakening of these specific NIST curves far less likely to be true
The research does not mean that all aspects of the original selection are fully understood, but Anthropic pointed out that some unknowns remain around the selection of certain base points and the reasons for skipping certain mathematically viable candidate curves
The breakthrough shows that artificial intelligence may be a tool that researchers can use to challenge long-held beliefs about widely used cryptographic systems
