A digital signature scheme known as HAWK, which was under consideration for official US standards as part of post-quantum cryptographic algorithms, has been put out of commission after an AI security model called Mythos revealed a critical flaw. Despite surviving two rounds of testing by NIST (the National Institute of Standards and Technology), the algorithm is now withdrawn following the findings.

The results from Anthropic’s Mythos are significant as they expose methods that could potentially reduce the work required to break cryptographic systems, even though these are weaker versions used for adversarial peer review. This highlights the continuing need for robustness in cryptographic algorithms against future quantum computing threats.

Even before this development, Anthropic was lauding its findings on two separate cryptographic problems: weaknesses in both HAWK and the widely used AES cipher. The outcomes, while incremental, could signal important advances in breaking cryptography that is crucial to privacy and security.

The challenge for developers of post-quantum cryptographic algorithms remains high as these systems need not only to survive rigorous testing but also withstand the evolving capabilities of AI and quantum computing technologies.