Privacy Policy Enforcement Guardrails for Data-Sensitive Retrieval-Augmented Generation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911735629217792 |
|---|---|
| author | Zafar, Osama Nemecek, Alexander Zhang, Yiqian Li, Wenbiao Ganguly, Debargha Singh, Vikash Chaudhary, Vipin Ayday, Erman |
| author_facet | Zafar, Osama Nemecek, Alexander Zhang, Yiqian Li, Wenbiao Ganguly, Debargha Singh, Vikash Chaudhary, Vipin Ayday, Erman |
| contents | Standard PII filters often miss contextual data leakage in RAG systems, such as non-regulated attribute clusters that collectively identify individuals. We introduce a Privacy Policy Enforcement (PPE) framework using dual one-class density estimators with fused text embeddings and a calibrated abstain region for out-of-distribution inputs. Using an axis-stratified, multi-LLM synthetic data pipeline across medicine, finance, and law, we found that traditional Gaussian Mixture baselines fail on borderline-safe stress tests by focusing on linguistic register rather than content.
Our proposed T3+OCSVM detector, trained on safe and borderline-safe data, achieves a borderline AUROC of 0.93+ while reducing false positives by 44-55 percentage points and maintaining millisecond latency. Compared to supervised MLP classifiers or 14B-parameter LLM judges, our framework offers superior operational suitability, as the former suffers from high abstention rates and the latter from latency and calibration issues. This methodology provides a robust stress-testing standard for any synthetic-data-trained classifier. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_17034 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Privacy Policy Enforcement Guardrails for Data-Sensitive Retrieval-Augmented Generation Zafar, Osama Nemecek, Alexander Zhang, Yiqian Li, Wenbiao Ganguly, Debargha Singh, Vikash Chaudhary, Vipin Ayday, Erman Machine Learning Artificial Intelligence Cryptography and Security Standard PII filters often miss contextual data leakage in RAG systems, such as non-regulated attribute clusters that collectively identify individuals. We introduce a Privacy Policy Enforcement (PPE) framework using dual one-class density estimators with fused text embeddings and a calibrated abstain region for out-of-distribution inputs. Using an axis-stratified, multi-LLM synthetic data pipeline across medicine, finance, and law, we found that traditional Gaussian Mixture baselines fail on borderline-safe stress tests by focusing on linguistic register rather than content. Our proposed T3+OCSVM detector, trained on safe and borderline-safe data, achieves a borderline AUROC of 0.93+ while reducing false positives by 44-55 percentage points and maintaining millisecond latency. Compared to supervised MLP classifiers or 14B-parameter LLM judges, our framework offers superior operational suitability, as the former suffers from high abstention rates and the latter from latency and calibration issues. This methodology provides a robust stress-testing standard for any synthetic-data-trained classifier. |
| title | Privacy Policy Enforcement Guardrails for Data-Sensitive Retrieval-Augmented Generation |
| topic | Machine Learning Artificial Intelligence Cryptography and Security |
| url | https://arxiv.org/abs/2605.17034 |