The essay argues that intelligence agencies and aligned actors are treating AI systems’ data-ingestion process as a new domain for influence operations, since deliberately seeded text, images, and commentary can ’tilt the mirror’ that AI holds up to the world. It draws parallels to historical narrative-management campaigns around Vietnam, COVID-19, Afghanistan, Ukraine, and Wikipedia/bot-network editing, and describes a tech-intelligence feedback loop as newsrooms increasingly use AI to generate coverage. It calls for treating AI outputs as an angled mirror rather than a neutral or authoritative source, and for scrutinizing the data and interests shaping those outputs.
