Post #4171913

Journalists not giving technical details or linking research reports is a PITA. Not exactly the linked article, but I am currently working through this report, which benchmarks LLMs: »CryptanalysisBench: Can LLMs do Cryptanalysis?« by Lukas Fluri, Avital Shafran, Nicholas Carlini, Matthew Jagielski, Milad Nasr, Orr Dunkelman, Eyal Ronen and Florian Tramèr They come to the conclusion that LLMs can find vulnerabilities in the implementation of crypt algorithms, but not yet in the algorithms themselves. We will get a lot more CVEs in crypt libraries in the near future. https://arxiv.org/abs/2607.18538 CC: @SteveBellovin@infosec.exchange