Contextual Integrity

Contextual integrity (CI) focuses on ensuring information flows appropriately within specific social contexts, addressing privacy concerns in increasingly data-driven systems. Current research emphasizes evaluating and improving CI in large language models (LLMs) and AI assistants, particularly focusing on their ability to reason about and adhere to privacy norms during inference, using techniques like prompting strategies and benchmark datasets to assess performance. This work is crucial for developing trustworthy AI systems that align with user expectations regarding data privacy and responsible information sharing, impacting both the ethical development of AI and the design of privacy-preserving technologies.