Georgia Institute of Technology
Extracting Hidden, Protected, and Deleted User Interaction Contexts from Applications and AI Systems
Abstract
dc:description.abstractIn this dissertation, I make hidden, protected, and deleted interaction context measurable and actionable by extracting it from real software stacks. In DVa, I uncover victim apps and attribute victim-specific a11y abuse vectors by combining victim modeling with abuse-vector-guided symbolic analysis on real malware. In SOMBRA, I identify where app-side a11y protections fail to transfer to mobile websites by modeling and comparing the a11y information exposed by native apps and by browser-rendered counterparts across four different browsers. In ORISA, I recover and rehost deleted conversation context from local LLM apps by extracting and realigning batch tokens and KV cache state so investigators can interrogate a live session with recovered context.
Degree
thesis:*- Name thesis:degree_name
- Electrical and Computer Engineering, PhD
- Department thesis:degree_department
- Electrical and Computer Engineering
- Grantor
- Georgia Institute of Technology
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Xu, Haichuan
- Advisor dc:contributor.advisor
-
- Saltaformaggio, Brendan D.
- Committee members dc:contributor.committeemember
-
- Zonouz, Saman
- Specter, Michael
- Li, Frank
- Ozmen, Ozgur
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/81689