Researchers at Anthropic have used their Claude AI model to discover improved ways to attack cryptographic algorithms, including a significantly weakened digital signature scheme called HAWK and an improved attack on reduced-round AES, the most widely used symmetric cipher. These attacks do not currently affect any production systems but demonstrate the potential for frontier AI models to help discover flaws in important cryptographic algorithms. AI summary
Firehose
Filtered to Hacker News, tagged “Claude Shannon” · clear filters
Browse: People · Companies · Papers · Podcasts · Hacker News · Deep dives