Anthropic’s Claude Mythos AI reportedly finds new attacks on cryptographic algorithms

luisdans · x · 2026-07-29

WatcherGuru reports that Anthropic's Claude Mythos AI found new ways to attack cryptographic algorithms, including a post-quantum encryption candidate. The post treats this as a security/research result: an AI system is being used to uncover weaknesses in cryptographic schemes rather than just benchmark language ability.

Related event: Anthropic Unveils CryptanalysisBench as Claude Finds Crypto Flaws(7 posts)→

Original post →

More from Research

Research channel →