Anthropic said Claude Mythos Preview helped derive a key-recovery attack against HAWK-256 by exploiting a hidden lattice symmetry, and the company released reproducibility code while stressing that the result does not affect production systems. It also reported a 200- to 800-fold speedup for an attack on seven-round AES-128, but said the attack still requires impractical chosen plaintexts and does not threaten real-world deployments. #HAWK-256 #AES-128 #ClaudeMythosPreview #NIST
Keypoints
- Anthropic used Claude Mythos Preview to find a key-recovery attack against HAWK-256.
- The HAWK attack exploits a previously unused lattice symmetry.
- The released implementation can recover a usable HAWK-256 signing key in hours on a 96-core server.
- Anthropic also found a faster attack on seven-round AES-128 by removing a 256-way guessing step.
- Both results remain impractical for production systems and do not break full-strength targets.
Read More: https://thehackernews.com/2026/07/claude-ai-just-cracked-post-quantum.html