Research

Paper

AI LLM February 26, 2026

IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computation

Authors

Yanpei Guo, Wenjie Qu, Linyu Wu, Shengfang Zhai, Lionel Z. Wang, Ming Xu, Yue Liu, Binhang Yuan, Dawn Song, Jiaheng Zhang

Abstract

Commercial large language models are typically deployed as black-box API services, requiring users to trust providers to execute inference correctly and report token usage honestly. We present IMMACULATE, a practical auditing framework that detects economically motivated deviations-such as model substitution, quantization abuse, and token overbilling-without trusted hardware or access to model internals. IMMACULATE selectively audits a small fraction of requests using verifiable computation, achieving strong detection guarantees while amortizing cryptographic overhead. Experiments on dense and MoE models show that IMMACULATE reliably distinguishes benign and malicious executions with under 1% throughput overhead. Our code is published at https://github.com/guo-yanpei/Immaculate.

Metadata

arXiv ID: 2602.22700
Provider: ARXIV
Primary Category: cs.CR
Published: 2026-02-26
Fetched: 2026-02-27 04:35

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.22700v1</id>\n    <title>IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computation</title>\n    <updated>2026-02-26T07:21:02Z</updated>\n    <link href='https://arxiv.org/abs/2602.22700v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.22700v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Commercial large language models are typically deployed as black-box API services, requiring users to trust providers to execute inference correctly and report token usage honestly. We present IMMACULATE, a practical auditing framework that detects economically motivated deviations-such as model substitution, quantization abuse, and token overbilling-without trusted hardware or access to model internals. IMMACULATE selectively audits a small fraction of requests using verifiable computation, achieving strong detection guarantees while amortizing cryptographic overhead. Experiments on dense and MoE models show that IMMACULATE reliably distinguishes benign and malicious executions with under 1% throughput overhead. Our code is published at https://github.com/guo-yanpei/Immaculate.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CR'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-02-26T07:21:02Z</published>\n    <arxiv:primary_category term='cs.CR'/>\n    <author>\n      <name>Yanpei Guo</name>\n    </author>\n    <author>\n      <name>Wenjie Qu</name>\n    </author>\n    <author>\n      <name>Linyu Wu</name>\n    </author>\n    <author>\n      <name>Shengfang Zhai</name>\n    </author>\n    <author>\n      <name>Lionel Z. Wang</name>\n    </author>\n    <author>\n      <name>Ming Xu</name>\n    </author>\n    <author>\n      <name>Yue Liu</name>\n    </author>\n    <author>\n      <name>Binhang Yuan</name>\n    </author>\n    <author>\n      <name>Dawn Song</name>\n    </author>\n    <author>\n      <name>Jiaheng Zhang</name>\n    </author>\n  </entry>"
}