South Korea certified a hybrid post-quantum encryption module as an AI model broke the HAWK algorithm in 60 hours. Two days before South Korean cybersecurity firm ITCEN PNS announced its cryptographic library EdgeCrypto v4.2 had cleared the country's mandatory government certification program, an AI model built by Anthropic discovered a fatal flaw in HAWK, an algorithm being considered as a future standard. By July 29, the day ITCEN PNS publicized its certification, HAWK's developers had withdrawn it from NIST's post-quantum standardization process. EdgeCrypto had been built precisely so that it would not need HAWK inside it. Anthropic's Claude Mythos Preview model, using a scaffold built on Claude Code, worked semi-autonomously for approximately 60 hours and found a mathematical symmetry—a nontrivial automorphism in HAWK's lattice structure—that every prior human reviewer had missed. The attack effectively halves the algorithm's security level, meaning HAWK would need to double its key sizes. Google post-quantum cryptography expert Sophie Schmieg said the result effectively ended HAWK's chances of becoming a standard. NIST emphasized the attack does not affect the three finalized post-quantum cryptographic standards finalized in August 2024.