Privacy Audits for Clinical Large Language Models

Florent Pollet Tong Wang Rahul Gupta Noémie Elhadad G. ̈ursoy K. Pollet T. Nikitin R. Wang N. Gupta G. El-Hadad G. ̈ursoy

· 0 citations · 40 references

TL;DR

Verified Extraction is introduced, an auditing framework that distinguishes identifiers attributable to fine-tuning data from spurious or prior-driven outputs and quantifies recoverable leakage under explicit query budgets.

Semantic Scholar

View source

Similar papers

Aug 2026

Scalable privacy risk assessment of medical data processed by large language models

This paper empirically evaluates ChatGPT 3.5 and 4.0 using over 23,000 real user-generated medical queries, assessing their susceptibility to privacy breaches through quasi-identifiers such as age, location, phone number and national registration number and proposes a scalable privacy evaluation model that combines k-a...

Foad Jalali, Mehran Alidoost Nia · Journal of Supercomputing · 0 citations

Preprint Aug 2026

Making Clinical Language Models Auditable: Concept-Guided Fine-Tuning for Robust Prediction

CAST (Concept-guided Artifact Suppression Tuning), an SAE-based framework for auditable clinical text classification, improves over its corresponding fine-tuned encoder baselines and remains competitive with strong LLM baselines, while producing a feature-level audit trail of the clinical concepts that support each pre...

Jin Mu, Guan-Hua Chen · 0 citations

Open access Jul 2026

Do Open-Weight LLMs Respect Minimum-Necessary Access in Text-to-SQL? An Automated Audit on EHR Benchmarks

This paper presents the first empirical audit of open-weight LLMs on this access-scope dimension, using the EHRSQL 2024 benchmark over MIMIC-IV, and studies default model behaviour under ordinary, benign queries.

Alberto Ballesteros-Rodríguez, Lino González-García, Miguel-Ángel Sicilia et al. · Electronics · 0 citations

Conference Aug 2026

Improving Clinical Note De-identification via Post-NER Verification and Candidate Expansion

De-identification of clinical notes is critical for protecting patient privacy, yet existing approaches often struggle under real-world variation and provide limited support for auditing and error analysis. By examining the outputs of NER-based systems, we observe three recurring failure modes—false positives, false ne...

Ze Zhou, Ruo-Qian Zhang, Zhicheng Jiao et al. · IEEE/ACM International Confe... · 0 citations

Jul 2026

Privacy Leakage in Federated Learning in Radiology Reports: A Comparative Evaluation of Tokenizer-Driven Privacy Risks

Substantial portions of report text are recoverable from FL gradients even at larger batch sizes and with domain-specific tokenizers, and safeguards such as secure aggregation and differential privacy are likely necessary to meet HIPAA and GDPR requirements for FL in radiology NLP.

Santhosh Parampottupadam, Andres Martinez, D. Bounias et al. · arXiv.org · 0 citations

#small language model Preprint Aug 2026

Future Querying: Can LLMs Serve as Implicit Medical World Models?

This work introduces future querying, a paradigm that probes whether large language models can function as implicit medical world models by evaluating their ability to answer time-indexed clinical queries about a patient's future, and shows that small, locally fine-tuned open-weight models can match or approach larger...

Siri Willems, James Butterworth, L. Goetschalckx et al. · 0 citations