Anthropic’s AI model, Claude, has achieved a remarkable feat by solving a cryptographic puzzle that has puzzled experts for 25 years. The mystery revolved around the seeds used in NIST’s standard curves for digital signatures, which were provided by the NSA without clear explanation. This lack of transparency has led to ongoing suspicions regarding potential backdoors embedded within the cryptographic standards.
In an impressive demonstration of its capabilities, Claude deciphered the origins of these seeds in under 30 minutes, identifying them as stemming from the ‘lrand48’ function, a simple C random number generator. This finding significantly diminishes the likelihood of deliberate backdoors being present in the cryptographic standards, providing a clearer understanding of their integrity.
This breakthrough not only showcases the advanced problem-solving abilities of AI models like Claude but also has implications for the trustworthiness of digital security protocols. As AI continues to evolve, its role in addressing complex challenges in cryptography and cybersecurity is becoming increasingly vital.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
