An Anthropic Claude AI Model Finds Flaws in Tough-to-Crack Encryption Algorithms

​Claude Mythos Preview discovered new attacks in testing against weakened cryptographic algorithms, which protect online financial transactions, private communications and more.