AI Model Finds New Weaknesses in Cryptographic Algorithms
Anthropic's AI model, Claude Mythos Preview, has identified new weaknesses in cryptographic algorithms HAWK and AES. The findings show that frontier AI models can detect mathematical flaws that human reviewers have missed for years.
Claude discovered a previously unused symmetry in HAWK, enabling a faster key recovery attack that effectively halves the scheme's key strength. In the smallest HAWK 256 configuration, the estimated cost of recovering a key fell from 2 to the power of 64 operations to 2 to the power of 38.
The company estimates that HAWK would need to double its proposed key sizes to maintain the intended security level, reducing its efficiency advantages. Claude developed this attack in about 60 hours using a multi-agent research system with access to mathematical software and computational tools.