AC-LoRA: (Almost) Training-Free Access Control-Aware Multi-Modal LLMs

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lazier, Lara Magdalena, Dhar, Aritra, Stambolic, Vasilije, Cavigelli, Lukas
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915537675616256
author Lazier, Lara Magdalena
Dhar, Aritra
Stambolic, Vasilije
Cavigelli, Lukas
author_facet Lazier, Lara Magdalena
Dhar, Aritra
Stambolic, Vasilije
Cavigelli, Lukas
contents Corporate LLMs are gaining traction for efficient knowledge dissemination and management within organizations. However, as current LLMs are vulnerable to leaking sensitive information, it has proven difficult to apply them in settings where strict access control is necessary. To this end, we design AC-LoRA, an end-to-end system for access control-aware corporate LLM chatbots that maintains a strong information isolation guarantee. AC-LoRA maintains separate LoRA adapters for permissioned datasets, along with the document embedding they are finetuned on. AC-LoRA retrieves a precise set of LoRA adapters based on the similarity score with the user query and their permission. This similarity score is later used to merge the responses if more than one LoRA is retrieved, without requiring any additional training for LoRA routing. We provide an end-to-end prototype of AC-LoRA, evaluate it on two datasets, and show that AC-LoRA matches or even exceeds the performance of state-of-the-art LoRA mixing techniques while providing strong isolation guarantees. Furthermore, we show that AC-LoRA design can be directly applied to different modalities.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11557
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AC-LoRA: (Almost) Training-Free Access Control-Aware Multi-Modal LLMs
Lazier, Lara Magdalena
Dhar, Aritra
Stambolic, Vasilije
Cavigelli, Lukas
Cryptography and Security
Artificial Intelligence
Corporate LLMs are gaining traction for efficient knowledge dissemination and management within organizations. However, as current LLMs are vulnerable to leaking sensitive information, it has proven difficult to apply them in settings where strict access control is necessary. To this end, we design AC-LoRA, an end-to-end system for access control-aware corporate LLM chatbots that maintains a strong information isolation guarantee. AC-LoRA maintains separate LoRA adapters for permissioned datasets, along with the document embedding they are finetuned on. AC-LoRA retrieves a precise set of LoRA adapters based on the similarity score with the user query and their permission. This similarity score is later used to merge the responses if more than one LoRA is retrieved, without requiring any additional training for LoRA routing. We provide an end-to-end prototype of AC-LoRA, evaluate it on two datasets, and show that AC-LoRA matches or even exceeds the performance of state-of-the-art LoRA mixing techniques while providing strong isolation guarantees. Furthermore, we show that AC-LoRA design can be directly applied to different modalities.
title AC-LoRA: (Almost) Training-Free Access Control-Aware Multi-Modal LLMs
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2505.11557