Anthropic's Mythos Identifies Vulnerabilities in HAWK Post-Quantum Scheme

Anthropic’s Claude Mythos model identified vulnerabilities in the HAWK post-quantum signature scheme after 60 hours of analysis, costing $100,000 in API usage. Human experts had reviewed HAWK for over two years prior to this analysis. Anthropic stated that while the findings do not affect current systems, they illustrate how AI could challenge fundamental assumptions in internet security. This follows Anthropic’s recent advancements in AI model performance and revenue growth, highlighting the ongoing intersection of AI and cybersecurity challenges. For practitioners, the results suggest a need to reassess the robustness of cryptographic algorithms in light of AI capabilities, as noted by The Decoder.