LatentAudit: Real-Time White-Box Faithfulness Monitoring for Retrieval-Augmented Generation with Verifiable Deployment

TL;DR

LatentAudit is a real-time, white-box monitoring method for RAG systems that assesses faithfulness using residual-stream geometry, requiring no extra judge model and enabling public verification.

Contribution

Introduces LatentAudit, a simple quadratic rule that measures residual-stream activations to verify answer support in RAG systems without auxiliary models.

Findings

Achieves 0.942 AUROC on PubMedQA with minimal overhead.

Remains stable across multiple models and benchmarks.

Enables verifiable deployment with 99.8% AUROC preservation at 16-bit precision.

Abstract

Retrieval-augmented generation (RAG) mitigates hallucination but does not eliminate it: a deployed system must still decide, at inference time, whether its answer is actually supported by the retrieved evidence. We introduce LatentAudit, a white-box auditor that pools mid-to-late residual-stream activations from an open-weight generator and measures their Mahalanobis distance to the evidence representation. The resulting quadratic rule requires no auxiliary judge model, runs at generation time, and is simple enough to calibrate on a small held-out set. We show that residual-stream geometry carries a usable faithfulness signal, that this signal survives architecture changes and realistic retrieval failures, and that the same rule remains amenable to public verification. On PubMedQA with Llama-3-8B, LatentAudit reaches 0.942 AUROC with 0.77,ms overhead. Across three QA benchmarks and five model families (Llama-2/3, Qwen-2.5/3, Mistral), the monitor remains stable; under a four-way stress test with contradictions, retrieval misses, and partial-support noise, it reaches 0.9566--0.9815 AUROC on PubMedQA and 0.9142--0.9315 on HotpotQA. At 16-bit fixed-point precision, the audit rule preserves 99.8% of the FP16 AUROC, enabling Groth16-based public verification without revealing model weights or activations. Together, these results position residual-stream geometry as a practical basis for real-time RAG faithfulness monitoring and optional verifiable deployment.