CoCo Seminar: "New Gatekeepers: News Source Citation Patterns in AI-Powered Search" by Kai-Cheng Yang (Computing, Binghamton University)

CoCo Seminar Series
Fall 2025

New Gatekeepers: News Source Citation Patterns in AI-Powered Search

Dr. Kai-Cheng Yang, Assistant Professor, School of Computing, Binghamton University

Wednesday September 17, 2025 12:15-1:15pm EDT
Hybrid (EB-T1 & Zoom; meeting link below)
https://binghamton.zoom.us/j/96780930614?pwd=aoU62aBXbMcoSgDOG0knIUvWFL9xLe.1

Abstract:
AI-powered search systems that integrate large language models (LLMs) are rapidly reshaping how people access news and information, acting as new gatekeepers of the online information ecosystem. Yet, little is known about the news citation patterns underlying these systems. In this talk, I will present our study of responses from multiple AI search systems to both generic queries and fact-checking requests. We find that news citations are highly concentrated among a small set of outlets and display a marked liberal bias, though low-credibility sources are rarely cited. To probe the origins of these patterns, we directly audit how LLMs evaluate news sources. We show that while LLMs consistently recognize high-credibility outlets, they systematically rate left-leaning sources as more credible. These findings highlight emerging challenges for fairness, diversity, and credibility in AI-mediated search. I conclude by discussing implications for societal trust, system design, and governance.

Speaker bio:
Kai-Cheng Yang is an Assistant Professor in the School of Computing at Binghamton University. He obtained his Ph.D. in Informatics from Indiana University. Before joining Binghamton University, he spent two years as a postdoctoral researcher at Northeastern University. His research focuses on creating safe, fair, and trustworthy online information platforms by identifying how malicious actors and flawed systems distort information flow and developing effective countermeasures. His work spans social bots, misinformation, and algorithmic biases. Currently, he is exploring how generative AI is being misused in these contexts and how to harness this technology to protect against these threats.

For more information, contact Hiroki Sayama (sayama@binghamton.edu).
http://coco.binghamton.edu/

Binghamton Center of Complex Systems | Website | View More Events

Contact the organizers