Scaling Security Understanding: AI-Driven Conversational Interfaces for Security and Privacy Reports of Thousands Apps
| Autor | Heid, Kris; Heider, Jens |
|---|---|
| Datum | 2026 |
| Art | Journal Article |
| Abstrakt | This paper presents a conversational AI system that transforms large-scale mobile app security and privacy reports into interactive, user-friendly dialogues. Enterprises and institutions must ensure that apps on company-managed devices comply with internal security and privacy standards, often relying on services that perform regular, automated assessments of hundreds to thousands of iOS and Android apps. The resulting PDF reports can exceed 10,000 pages, making it difficult for security officers and other stakeholders to gain an overview and identify critical issues. Our system introduces customer-specific AI agents that securely access and index each customer's reports, enabling users to query findings, clarify test methodologies, and explore issues in natural language. We outline the architecture, lessons learned, and evaluation proposal for tenant-isolated access, large-scale retrieval and summarization, and dialogue design for follow-up questions and varying technical depth. |
| Konferenz | International Conference on Mobile Systems and Pervasive Computing 2026 |
| Url | https://publica.fraunhofer.de/handle/publica/526166 |
Schwachstellen und Sicherheitslücken
Aus- und Weiterbildung
Informieren Sie sich über die Aus- und Weiterbildungsangebote der in ATHENE mitwirkenden Institutionen