The goal of the BENGAL targeted super seedling is to understand LLM threat modes, quantify them and to find novel methods to correct threats and vulnerabilities or to work resiliently with imperfect models. IARPA seeks to develop and incorporate novel technologies to efficiently probe large language models to detect and characterize LLM threat modes and vulnerabilities. Performers will focus on one or more of the topic domains below, clearly articulate a taxonomy of threat modes within their domain of interest and develop technologies to serve as an analog to ‘virus scan’ software.
BENGAL BAA
November 17, 2023
