Searcharxiv⌕ Search

arXiv subjects

Maxime Puys

Publications and source records attributed to Maxime Puys.

3 recordsLinked to original sources

Isolation Is Not Confidentiality: Security Boundaries of Geometric Access Control in RAG

Retrieval-Augmented Generation (RAG) systems increasingly rely on vector databases containing documents governed by different access-control policies. Recent search-preserving geometric approaches enforce such access control by structuring the embedding space so that transformed embeddings, which we call protected representations, remain retrievable without exposing information across policy boundaries. We provide a unified security analysis of this paradigm, covering policy isolation, representation confidentiality, query forgery resistance, and policy-membership hiding. We consider both a curious cloud service provider observing the protected representations stored in the vector database and a curious user holding aligned pairs of original embeddings and protected representations. Evaluating the available implementations of the works proposing these geometric transformation mechanisms across three encoders, we find that geometric transformations preserve retrieval utility and provide strong policy isolation, but do not provide the other security properties. A curious user can recover the original embeddings from a few hundred pairs and forge policy-compatible queries at a higher cost that grows with encoder dimension for secret per-policy transformations. Moreover, policy-dependent geometry can leak policy membership even when unauthorized retrieval is prevented. These results show that policy isolation, confidentiality, forgery resistance, and membership hiding are distinct security objectives.

cs.CR↗

A Security Meta-Model for Retrieval-Augmented Generation Systems

Retrieval-Augmented Generation (RAG) systems extend large language models (LLMs) with external knowledge through a multi-stage pipeline. While this architecture can improve the factual grounding of generated answers, it introduces structural attack surfaces that extend beyond those of standalone LLMs. In this paper, we introduce a security meta-model that captures explicit causal relationships between RAG surfaces, attacks, weaknesses, risks, and CIA impact (Confidentiality, Integrity, Availability). Its purpose is to provide security engineers with a structured and user-friendly framework for gathering and assessing the risks, weaknesses, and mitigations relevant to their RAG deployment. We designed the meta-model through an iterative, structured analysis of 43~publications (2023--2026) and instantiated it as a catalog populated with the security threats and remediations reported in the literature. Filtering the catalog according to a deployment configuration produces a risk profile containing the risks applicable to that deployment. An interactive web visualizer lets users navigate the catalog as a graph, follow causal chains, and explore stakeholder-specific views. Analysis of the catalog revealed a persistent imbalance between attack-focused and defense-focused research, a concentration of threats at ingestion, and coverage gaps affecting output integrity. Coverage is assessed against the OWASP LLM Top~10, and operational applicability is illustrated across textual, graph-based, and multimodal RAG configurations.

cs.CR↗

Private Multi-party Matrix Multiplication and Trust Computations

This paper deals with distributed matrix multiplication. Each player owns only one row of both matrices and wishes to learn about one distinct row of the product matrix, without revealing its input to the other players. We first improve on a weighted average protocol, in order to securely compute a dot-product with a quadratic volume of communications and linear number of rounds. We also propose a protocol with five communication rounds, using a Paillier-like underlying homomorphic public key cryptosystem, which is secure in the semi-honest model or secure with high probability in the malicious adversary model. Using ProVerif, a cryptographic protocol verification tool, we are able to check the security of the protocol and provide a countermeasure for each attack found by the tool. We also give a randomization method to avoid collusion attacks. As an application, we show that this protocol enables a distributed and secure evaluation of trust relationships in a network, for a large class of trust evaluation schemes.

cs.CR↗