Anthropic's Claude Mythos model discovers new cryptanalytic attacks

Anthropic's Claude Mythos model identified a key-recovery attack on the HAWK post-quantum signature scheme, reducing HAWK-512 security from 128 to 108 bits. The model also produced an improved attack on 7-round AES-128, though it found no threats to the full 10-round AES standard.
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Modular web interface unifies voice synthesis tools
- Arkon serves internal docs to Claude via permission-scoped RAG
- OpenAI launches GPT-5.6 Sol for Plus and Pro ChatGPT users
- ScienceClaw is a personal research assistant with 1,900+ science tools
- Kokoro Web: open-source browser TTS powered by Kokoro-82M