CI-Bench: Benchmarking Contextual Integrity of AI Assistants on Synthetic Data

TL;DR

CI-Bench is a synthetic benchmark designed to evaluate AI assistants' ability to protect personal information during inference, using a scalable data pipeline and the Contextual Integrity framework to assess information flow across context dimensions.

Contribution

The paper introduces a novel, scalable synthetic data pipeline and a comprehensive benchmark for evaluating privacy-preserving capabilities of AI assistants based on Contextual Integrity.

Findings

Demonstrated the benchmark's ability to evaluate information flow

Showcased the need for careful training of AI assistants for privacy

Provided a baseline naive AI assistant for comparison

Abstract

Advances in generative AI point towards a new era of personalized applications that perform diverse tasks on behalf of users. While general AI assistants have yet to fully emerge, their potential to share personal data raises significant privacy challenges. This paper introduces CI-Bench, a comprehensive synthetic benchmark for evaluating the ability of AI assistants to protect personal information during model inference. Leveraging the Contextual Integrity framework, our benchmark enables systematic assessment of information flow across important context dimensions, including roles, information types, and transmission principles. We present a novel, scalable, multi-step synthetic data pipeline for generating natural communications, including dialogues and emails. Unlike previous work with smaller, narrowly focused evaluations, we present a novel, scalable, multi-step data pipeline that synthetically generates natural communications, including dialogues and emails, which we use to generate 44 thousand test samples across eight domains. Additionally, we formulate and evaluate a naive AI assistant to demonstrate the need for further study and careful training towards personal assistant tasks. We envision CI-Bench as a valuable tool for guiding future language model development, deployment, system design, and dataset construction, ultimately contributing to the development of AI assistants that align with users' privacy expectations.