Anthropic says Claude solved 25-year cryptography mystery in 30 minutes

Investing.com -- Anthropic said its Claude AI model solved a decades-old mystery surrounding the origins of cryptographic standards used to secure digital communications, tracing key numbers to a simple random-number generator seeded with timestamps.

The discovery eases long-standing concerns that the U.S. National Security Agency may have deliberately weakened some widely used cryptographic tools when their parameters were developed in the late 1990s. Anthropic said the findings do not reveal a security flaw and require no action from users.

The mystery centers on the Elliptic Curve Digital Signature Algorithm, or ECDSA, which helps verify the authenticity of digital messages, software and transactions. When the National Institute of Standards and Technology adopted standardized elliptic curves in 2000, the NSA had provided the underlying numbers without explaining how they were generated, fueling speculation about potential backdoors.

Claude cracked the puzzle in less than half an hour by analyzing the numbers for patterns rather than searching for a rumored secret phrase. The model identified lrand48, a random-number generator commonly found in older C programming libraries, and reconstructed its internal state.

The recovered starting points corresponded to Unix timestamps from 1997 and 1998. By reproducing the original process, Claude generated the published seed values exactly, providing strong evidence of how the numbers were created.

The model also traced the starting points of six older curves to a random-number generator in a mathematical software package developed by Arjen Lenstra.

Some questions remain unresolved. The researchers could not explain the origins of several other curve parameters, including one seed and the base points of four major NIST curves. They also found that the original process skipped some candidates that appeared capable of producing valid curves, although they found no evidence that the selections were deliberately weakened.

Anthropic said the findings make the theory of an intentional backdoor less plausible. The researc

Original Article

You May Also Be Interested In

• Nscale in talks for $45 billion Anthropic AI deal as IPO looms
• Rapport Therapeutics soars over 50% as bipolar mania drug meets trial goals
• Alvotech shares jump premarket after FDA approves new Simlandi process