As large language models integrate into daily workflows—from personal assistants to workplace tools—they handle sensitive information from multiple sources yet struggle to reason about what to share, with whom, and when. In this talk, we explore critical gaps in LLMs' privacy reasoning through complementary benchmarks.

Having just completed her PhD in Computer Science at Brown University (advised by Professor Suresh Venkatasubramanian), Pegah Nokhiz is a Postdoctoral Fellow at the Digital Life Initiative, Cornell Tech. She was an affiliate of Brown's Center for Technological Responsibility, Reimagination and Redesign (CNTR) at the Data Science Institute.

The Digital Life Seminar series offers students and guests an opportunity to engage actively with leading scholars and practitioners researching and responding to the development and application of digital technologies.

AI governance underscores the AI “supply chain” and the “many hands” involved in the production of AI systems. Even when it addresses downstream risks, the focus remains on AI's producers. Although invaluable and important, this production-centered approach risks overlooking what happens when real people use AI tools. This paper identifies and theorizes AI governance’s missing half, which I call the “demand side.”

The rise of large language models (LLMs) is reshaping knowledge production, raising urgent questions for research communication and publishing writ large. Drawing on qualitative survey responses from over 850 academic book authors from across a range of fields and institutions, we highlight widespread concern about the unlicensed use of in-copyright scientific and scholarly publications for AI training. Most authors are not opposed to generative AI, but they strongly favor consent, attribution, and compensation as conditions for use of their work.