The Privacy and Security Risks of Edge LLM Optimization
Edge deployment of Large Language Models requires optimization techniques that reduce memory, computation, and energy costs. However, preserving task performance does not necessarily preserve security. We argue that optimization is an information transformation that can alter representations, decision pathways, and trust boundaries, creating security drift and optimization-induced attack surfaces.
Authors
- Ren-Yi Huang (ORCID: https://orcid.org/0009-0006-8341-5835)
- Dumindu Samaraweera (ORCID: https://orcid.org/0000-0003-4097-5585)
- Prashant Shekhar (ORCID: https://orcid.org/0000-0003-2353-6740)
- Mingchen Li
- J. Morris Chang
Publication Details
- Journal
- arXiv (Cornell University)
- Published
- 2026-10-29
- DOI
- https://doi.org/10.48550/arxiv.2610.40543
- Primary Topic
- Adversarial Robustness in Machine Learning
- Type
- preprint