Anthropic's Claude AI model has resolved a decades-old cryptographic mystery regarding elliptic curves adopted by the US National Institute of Standards and Technology (NIST) in 2000. For years, researchers suspected that the NSA-provided starting numbers for these curves could contain hidden backdoors. Claude successfully traced five sets of these numbers to a standard computer program that utilized clock times from 1997 and 1998.
The AI completed this analysis in under 30 minutes, allowing anyone to reproduce the results. Anthropic stated that the findings increase confidence that the curves were selected without intentional weaknesses. The project earned a $36,864 bounty, which the company plans to donate to charity. Despite the discovery, some questions persist regarding why specific numbers were omitted during the original process.
The findings have reignited discussions about Bitcoin's security design. Bitcoin utilizes the secp256k1 curve, which relies on simple, publicly understood numbers. While some industry figures questioned the necessity of avoiding NSA-approved curves, others defended the choice as a precaution against potential manipulation. Claude identified no weaknesses in Bitcoin, and its underlying cryptography remains secure against current practical attacks.
Industry experts, including Blockstream CEO Adam Back, noted that the controversy stems from the NSA's failure to provide reproducible steps for its curve generation. While the AI discovery mitigates concerns about deliberate manipulation, long-term security discussions continue to focus on the potential threat of quantum computing to both Bitcoin and traditional cryptographic systems.
