Artificial intelligence has achieved what leading cryptography experts could not for two years. Anthropic's Claude model, after 60 hours of work, uncovered hidden vulnerabilities in two encryption systems—HAWK and a simplified version of AES. This is not about breaking current standards, but rather a warning signal for the future of digital security.

A Blow to the Post-Quantum Signature HAWK

The HAWK system was positioned as a digital signature resistant to attacks from quantum computers. For two years, experts found no flaws in it. However, Claude discovered a mathematical pattern that allows recovering the smallest HAWK key in just 238 steps instead of the expected 264. This means a reduction in computational complexity by approximately 67 million times. There is a solution—doubling the key size—but this eliminates HAWK's main advantage over its competitors.

Attack on AES: The "Möbius Bridge"

Claude also analyzed a lightweight version of AES with 7 rounds instead of the standard 10. The AI found a non-standard attack method, dubbed the "Möbius Bridge," which speeds up brute-forcing by 200–800 times. It took three days of the model working alone to find it, after which two specialists spent nearly a month double-checking the result.

Expert Opinion from Cryptalist

This case is not just a demonstration of AI capabilities. It raises a fundamental question: are we ready for future encryption standards to be broken not by humans, but by machines, and faster than we can implement them? For the cryptocurrency market, this is a signal that the race for post-quantum security could become even more intense than anticipated. While Bitcoin and Ethereum are not yet affected, in the coming years, NIST and Google will have to consider this new reality when selecting algorithms.