Reproduce “LLM-Redactor: An Empirical Evaluation of Eight Techniques for Privacy-Preserving LLM Requests” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Justice Owusu Agyemang, Jerry John Kponyo, Elliot Amponsah, Godfred Manu Addo Boakye, Kwame Opuni-Boachie Obour Agyekum
The authors evaluate a local shim that limits information sent from coding agents to cloud models. It routes answerable prompts locally, redacts detected spans, and rephrases requests before forwarding them. Across synthetic workloads, the combined approach has 0.6% combined leak and zero exact leaks across 500 PII prompts, but 31.3% combined leak on proprietary-code prompts; the study does not measure real-world privacy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.