Claude Mythos Finds New Weaknesses in Cryptographic Algorithms

Anthropic researchers have used Claude Mythos Preview to find improved ways to attack cryptographic algorithms, the mathematical methods used to secure online data. The first attack significantly weakens HAWK, a post-quantum digital signature scheme under consideration by NIST. Despite HAWK surviving two rounds of expert human review over two years, Claude Mythos improved the best-known attack on it in just 60 hours of work, effectively cutting its key strength in half.

The second result targets the Advanced Encryption Standard (AES), the most widely used symmetric cipher adopted by NIST in 2001. Claude Mythos identified a new way to attack round-reduced AES, adding to the body of cryptanalytic research on this foundational algorithm. These are substantial research advances, but they do not currently affect any production systems in use today.

These findings demonstrate a dramatic acceleration in cryptanalytic capability enabled by frontier AI models. Where human cryptanalysts spend years developing attack refinements, Claude Mythos achieved breakthroughs in days. As AI models continue to advance, the cryptographic community will need to adapt standards and key sizes to account for this new capability, much as the field previously adapted to the threat of quantum computers.