Anthropic says Claude helped researchers find weaknesses in cryptographic algorithms

teortaxesTex · x · 2026-07-29

Anthropic says Claude found weaknesses in cryptographic algorithms

Anthropic shared new research showing that Claude Mythos Preview helped its researchers discover weaknesses in cryptographic algorithms — the mathematical methods used to protect data privacy.

The accompanying discussion on X notes that the result appears to be a real improvement over the state of the art, even if the showcased AES-related attack is heavily simplified for demonstration purposes. The thread frames it as an example of Claude contributing to security research rather than just generating text.

Related event: Claude Discovers New Cryptographic Weaknesses in HAWK and Reduced-round AES(32 posts)→

Original post →

More from Research

Research channel →