Reproduce “A Generative Model of Contextual Integrity: Appropriate vs. Inappropriate Information Sharing” 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

Omer Ebead, Claude Formanek, Joel Z. Leibo

The authors simulate privacy decisions through conversations between two LLM actors: one holds private information, while the other requests it. The holder can question the request and respond to explanations. The framework scores sharing against contextual norms, where appropriateness depends on context and recipient entitlement. In baseline runs, appropriate withholding occurred in 40.7% (11/27) of Type III conversations, versus appropriate sharing in 100% (27/27) of Type IV conversations. These findings come from simulated English text conversations, not human behavior or deployed systems.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.