Stephen Meisenbacher, Alexandra Klymenko, Alexander Karpp, Florian Matthes. Proceedings of the Sixth Workshop on Privacy in Natural Language Processing. 2025.
Privacy-Enhancing Technologies (PETs) have been receiving rising attention in the academic sphere. In practice, however, the adoption of such technologies remains low. Beyond the implementation of a PET, it is not clear where along the process of software engineering PETs should be considered, and which activities must take place to facilitate their implementation. We investigate PETs in the software engineering process from the perspective of privacy requirements engineering. We conduct a systematic literature review and interview 10 privacy experts, exploring the integration of PETs into software engineering, as well as identifying associated challenges and their potential solutions. We also conduct a survey study with five experts and 100 developers to evaluate our created artifact. We propose a unified process diagram that illustrates the roles and activities involved in the implementation of PETs in software systems. We map the identified solution concepts to the diagram, highlighting vital stages of the software engineering process in tackling challenges in adopting PETs. The results of our survey study indicate general overall agreement with the presentation and content of the process diagram. Integrating PETs into software engineering involves a diverse cohort of roles and several challenges, but also concrete solution concepts that can be pinpointed in our process diagram. An analysis of available solutions juxtaposed with perceived challenges from privacy experts reveals there is much work to be done in increasing practitioner education on PETs.
The continued promise of Large Language Models (LLMs), particularly in their natural language understanding and generation capabilities, has driven a rapidly increasing interest in identifying and developing LLM use cases. In an effort to complement the ingrained "knowledge" of LLMs, Retrieval-Augmented Generation (RAG) techniques have become widely popular. At its core, RAG involves the coupling of LLMs with domain-specific knowledge bases, whereby the generation of a response to a user question is augmented with contextual and up-to-date information. The proliferation of RAG has sparked concerns about data privacy, particularly with the inherent risks that arise when leveraging databases with potentially sensitive information. Numerous recent works have explored various aspects of privacy risks in RAG systems, from adversarial attacks to proposed mitigations. With the goal of surveying and unifying these works, we ask one simple question: What are the privacy risks in RAG, and how can they be measured and mitigated? To answer this question, we conduct a systematic literature review of RAG works addressing privacy, and we systematize our findings into a comprehensive set of privacy risks, mitigation techniques, and evaluation strategies. We supplement these findings with two primary artifacts: a Taxonomy of RAG Privacy Risks and a RAG Privacy Process Diagram. Our work contributes to the study of privacy in RAG not only by conducting the first systematization of risks and mitigations, but also by uncovering important considerations when mitigating privacy risks in RAG systems and assessing the current maturity of proposed mitigations.