AI & Journalism: Who Controls the Narrative

Propaganda once required a state budget and a printing press. It now requires a prompt. Synthetic video, fabricated atrocity footage, and coordinated inauthentic accounts move faster than any newsroom can verify, and the mere existence of deepfakes lets authentic evidence be dismissed as fake. Meanwhile the economics that funded verification are collapsing: answer engines summarize a war or an election for millions without sending a reader to the outlet that paid to report it, newsrooms cut the desks that check claims, and a handful of model companies inherit the position of default narrator. Those companies are not neutral. They answer to advertisers, licensing partners, investors, and the governments that regulate them, and their training data, safety tuning, and content deals quietly encode which framings count as reasonable and which sources count as credible. When a model trained largely on Western media becomes the world's summarizer of Gaza, Sudan, or Kashmir, editorial judgment moves from accountable institutions to opaque corporate incentive. This panel asks what journalists still do that no system can replace, and who is left to hold power to account when the intermediary is owned by the powerful.