Shijing He "Exploring GenAI-Mediated Relationship-Level Coercion Social Engineering"
This talk can be viewed via zoom. (Note: this talk will be recorded)
Title: Exploring GenAI-Mediated Relationship-Level Coercion Social Engineering
Short abstract: Generative AI (GenAI) is reshaping social engineering by making manipulation more relational and coercive. Rather than relying on a single deceptive message, attackers can now cultivate trust over time, generate persuasive synthetic evidence, and exploit ongoing interpersonal dynamics (e.g., dependency, fear). This project investigates GenAI-mediated relationship-level coercion in human-AI companionship contexts, where high disclosure, persistent interaction histories, and limited transparency around data retention may increase user vulnerability. It focuses in particular on situations of constrained control, where individuals cannot safely rely on standard protective actions such as blocking and reporting. The project addresses four questions: how relational coercion unfolds across its lifecycle; how cross-platform infrastructures amplify attacker leverage; what protections remain feasible under constrained control; and how responsibilities should be distributed across users, platforms, and institutions. The project will contribute new conceptual and practical foundations for usable privacy and security by making relational coercion legible as a workflow and by producing human-centered protections that remain implementable in real-world GenAI ecosystems.