For 25 years, one suspicion haunted the mathematics protecting much of the internet: had the NSA secretly built a way to crack it？ Now, an AI model may have weakened that theory. And Bitcoins creator is suddenly part of the argument.
On Friday, Anthropic revealed that its Claude AI model solved a mystery that had baffled cryptographers for decades—in less than 30 minutes.
Sponsored
Sponsored
What Did Claude Discover About the NSA？
The controversy concerned elliptic curves, mathematical rules used to verify digital signatures without revealing private keys.
In 2000, US standards body NIST adopted several curves containing starting numbers supplied by the NSA.
Nobody explained where those numbers came from. Researchers feared the agency might have deliberately selected them to create a hidden weakness, or backdoor.
Claude traced five sets of numbers to an ordinary computer program that generated random-looking values using clock times from 1997 and 1998.
Anyone can now reproduce the results.
Sponsored
Sponsored
“In fact, this makes us more confident the curves were chosen without any backdoors or intentional weaknesses,” Anthropic said.
The finding earned a $36,864 bounty, which will go to charity.
However, questions remain about why certain numbers were skipped. The discovery makes deliberate manipulation less likely, without ruling it out completely.
Was Satoshi Wrong to Avoid the NSA Curves？
Bitcoin uses a different curve, called secp256k1. Its design uses simple, publicly understood numbers, leaving little room to conceal suspicious choices.
Satoshi Nakamoto never publicly explained why he chose it.
After Edward Snowden‘s 2013 leaks exposed NSA efforts to weaken a separate encryption standard, Bitcoin’s approach gained credibility.
Ledger co-founder Nicolas Bacca reacted with “checkmate Satoshi,” arguing that the industry had avoided NSA curves in favor of a more complex alternative.
Blockstream CEO Adam Back pushed back, blaming the NSA for failing to explain its choices.
For Bitcoin holders, the practical answer is simple: nothing has changed.
Claude found no weakness in Bitcoin. Its underlying cryptography remains secure against known practical attacks.
The bigger long-term concern remains quantum computing, which could eventually threaten both Bitcoin and traditional cryptographic systems.