Abstract

This report describes efforts under DARPA's ASED program in the detection of social engineering attacks. We describe the signals used for detection, experiments with those signals, and the results from the programs shared testbed. The detection approached developed during the program fused weak signals from the metadata, linguistic content of an email, and a grounding of who a sender is in the external world. The report also describes the engineering integration efforts that were used to integrate into the program's shared testbed. In the program testbed, the system deployed under this effort achieved 99% performance on the program's testbed.

Date

2023

Authors

Marjorie Freedman, Genevieve Bartless