ORCID Number
0000-0003-0032-8201
Date of Award
Summer 7-10-2025
Access Type
Thesis - Open Access
Degree Name
Master of Science in Computer Science
Department
Electrical Engineering and Computer Science
Committee Chair
Shafika Showkat Moni
Committee Chair Email
monis@erau.edu
First Committee Member
Omar Ochoa
First Committee Member Email
Omar.Ochoa@erau.edu
Second Committee Member
Laxima Niure Kandel
Second Committee Member Email
Laxima.NiureKandel@erau.edu
College Dean
James W. Gregory
Abstract
Large-language models (LLMs) already power mission critical tasks such as command-and-control chat, satellite ground-station automation, military analytics, and cyber-defense. Since most of these services are offered through application programming interfaces (APIs) that still expose full or top-k logits and lack mature safeguards, they present a serious, often overlooked attack surface. Earlier work has shown how to rebuild the output projection layer or distill surface behavior, but no attack has produced a deployable clone within a tight query budget. In this thesis, we address this problem by presenting a practical pipeline for cloning LLMs under constrained settings. The approach first estimates the output projection matrix by collecting fewer than 10,000 top-$k$ logits from the target model and analyzing them with singular value decomposition (SVD). Next, it trains smaller student models with varying transformer depths on publicly available data to reproduce the original model’s internal patterns and outputs. Experiments show that a 6-layer student can match 97.6\% of the teacher model’s hidden-state geometry, with only a 7.31\% increase in perplexity and an NLL of 7.58. A lighter 4-layer version runs 17.1\% faster and uses 18.1\% fewer parameters, while maintaining strong performance. The entire attack completes in under 24 GPU hours without triggering API rate limits. These findings show that even adversaries with limited budgets can recreate the knowledge inside modern LLMs, highlighting the urgent need for stronger API protections and safer deployment practices.
Scholarly Commons Citation
Gharami, Kanchon, "In the Shadow of Prompts: Adversarial Attacks and Model Cloning in Large Language Models" (2025). Doctoral Dissertations and Master's Theses. 909.
https://commons.erau.edu/edt/909
Included in
Artificial Intelligence and Robotics Commons, Electrical and Computer Engineering Commons, Information Security Commons
