This study examines how researchers navigate information practices and ethical uncertainties when integrating generative AI into scholarly work. Through semi‑structured interviews with 18 U.S. researchers and a grounded theory analysis, we identify an “AI uncertainty gap” shaped by challenges in verifying reliability, provenance, and authorship boundaries. Participants adopt hybrid verification strategies that blend traditional literature searching with cross‑tool comparison. The study proposes an emergent model of AI‑mediated information behavior and offers implications for responsible AI use, academic integrity, and researcher training.
