Natural Language Engagement of Malicious Entities through a Social Interaction Service (NEMESIS)
Abstract
Project NEMESIS developed active engagement services designed to defend enterprise networks from sophisticated social engineering attack campaigns by adversaries that possess the resources to study potential victims, the skills to exploit this knowledge during dialog engagements with their victims, and the patience to defer a victims compromise for maximum eventual success and impact. Our team was directed by DARPA to spearhead the development of an Ensemble Dialog Management (EDM) system, capable of integrating multiple dialog generation strategies for the purpose of integration into a live defensive service. Our system demonstrated a 52 percent Rate of Engagement (RoE) against in the 4th release of this system and 93 percent in the 5th. In an experiment designed by Zetier and ARLIS to examine whether key administrators throughout SRI accurately follow corporate policies regarding the disclosure of information about previous employees to external third parties, we achieved a 32 percent RoE, comparable to what can be achieved by human testers, but scalable to much larger populations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 2022
- Accession Number
- AD1202408
Entities
People
- Phillip Porras
Organizations
- SRI International