SecureLLM: Using Inference-time Compositionality to Build Secure Language Models for Private, Sensitive, and Secret Data

ICLR 2026 Conference Submission19158 Authors

Keywords: Generative Models, Adapter Composition, Model Privacy, Model Security

TL;DR: We introduce an information security method for LLMs that relies on inference-time composition to protect private data without the need for probabilistic guardrails

Abstract: As Large Language Models (LLMs) increasingly support critical sectors such as healthcare, finance, and public governance, ensuring data confidentiality and robust access control is a pressing societal challenge. Traditional security mechanisms isolate sensitive resources from unauthorized users, yet existing LLM safety approaches often fail to enforce strict segregation of confidential data. In this work, we introduce SecureLLM, a novel compositional framework for building provably secure large language models (LLMs) that integrates fine-tuning with traditional access security measures to protect private information. By fine-tuning LLMs on segregated, “siloed” training data and composing their outputs at inference time based solely on a user’s verified credentials, SecureLLM not only prevents unauthorized data leakage but also enables accurate responses for complex queries spanning multiple data silos. Our method is demonstrated on a challenging natural-language-to-SQL translation task and is designed with real-world applications in mind—supporting sectors where protecting sensitive information is paramount.

Primary Area: generative models

Submission Number: 19158