Reproduce “FinancialAuditBench: Benchmark Construction under Differential Privacy Using Real-World Priors” 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
Jerry Huang, Sarvesh Babu, Matt Van Buren, Alexander Wang, Pranav Pillai, Arush Jain, James P. Burton, Julia Hockenmaier
FinancialAuditBench evaluates agents on synthetic audit workpaper tasks built from differentially private historical-audit patterns and auditor expertise. It contains 90 completion and review tasks across six engagements. Claude Opus 5.5 passed all applicable rubric checks on 69.17% of single-attempt runs, but all eight attempts passed on only 44.44% of tasks. Rubric passes do not establish error-free work; the benchmark covers selected audit tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.