What happens when the profession dedicated to finding truth collides with a technology that hallucinates? At a panel hosted by Princeton's Program in Journalism and the Center for Information Technology Policy, four journalists and researchers talked through the real effects of AI on newsrooms, from the New York Times to local papers covering 169 Connecticut towns.

Dylan Freedman, AI projects editor at the New York Times, describes how his team used AI to search 400 hours of leaked Zoom recordings before Election Day, and explains why a ChatGPT user in Toronto spent three weeks and over a million words believing he'd made a mathematical breakthrough. Hilke Schellmann, an investigative reporter and NYU professor, explains why fake whistleblowers with AI-generated documents are "a now problem" for journalism. Anjanette Delgado, group managing editor at Hearst Connecticut Media Group, describes using AI to pull comparable data from school budgets that arrive as everything from spreadsheets to PowerPoints with children's drawings. And Madelyne Xiao, a PhD candidate at Princeton and former New Yorker fact-checker, raises the question of what happens when the academic research journalists rely on as primary sources is itself potentially fabricated.

The conversation covers deepfakes, sycophantic chatbots, the security risks of feeding sensitive source material into AI tools, the collapse of activity on developer forums like Stack Overflow, and why one panelist bought a manual typewriter. Moderated by Eliza Griswold, director of Princeton's journalism program.

Recorded March 26, 2026, at the Chancellor Green Rotunda, Princeton University.