Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/149664
Title: Seguridad y privacidad ¿cómo valorar la seguridad sobre un sistema basado en IA?
Author: Garcia Berzosa, Hector
Tutor: Fernández García, Diego
Others: Isern-Deya, Andreu Pere  
Abstract: Currently, we are experiencing a boom related to Artificial Intelligence and its integration into systems and organizations of all sizes and sectors. Accompanying this growth, guidelines, methodologies, and protocols for secure development, maintenance, and implementation are continuously being developed, adapting to the ongoing changes in the technology. However, a lack has been observed in the analysis and formal communication of vulnerabilities. The purpose of this essay is to try to fill this gap. The author of this Master's Thesis has years of experience as a pentester and Threat Intelligence analyst, applying his approach and methodology to its development, aiming to approach the exercise from the perspective of a professional in Offensive Security. The document provides an overview of the current state of security in applications that use Artificial Intelligence. This contextualization is then leveraged to assist in generating a proposal for factors to consider when conducting a security analysis of such applications, in addition to proposing a model for presenting results. Finally, an in-depth analysis is carried out on vulnerabilities known as Prompt Injection, as it is particularly useful to illustrate the characteristics of vulnerabilities that affect Artificial Intelligence, especially those of the generative kind.
Keywords: artificial intelligence
vulnerability
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 7-Jan-2024
Publication license: http://creativecommons.org/licenses/by-nc-nd/3.0/es/  
Appears in Collections:Trabajos finales de carrera, trabajos de investigación, etc.

Files in This Item:
File Description SizeFormat 
hgarciaberTFM0124memoria.pdfMemoria del TFM1,63 MBAdobe PDFThumbnail
View/Open
Share:
Export:
View statistics

This item is licensed under aCreative Commons License Creative Commons