Anthropic says Claude found new attacks on HAWK and a reduced variant of AES

fkasummer · x · 2026-07-29

Anthropic says Claude Mythos Preview helped researchers discover new weaknesses in cryptographic algorithms.

The broader point of the post is that powerful language models may now help experts find cryptographic flaws faster, which raises both security concerns and research interest.

Related event: Anthropic's Claude Mythos Uncovers Flaws in Post-Quantum Crypto and AES(30 posts)→

Original post →

More from Research

Research channel →