How secondary coverage overtakes original government sources in AI interpretation. “Why does AI say the city issued a warning that officials never actually said?” A resident asks an AI assistant about a public safety advisory after a local incident. The response appears confident: it cites “city officials” and summarizes a restriction that was never formally issued. The wording matches a widely shared news article, not the original municipal release. The official update contained more precise language, including conditions and limitations, but that version is absent. The AI output is clear, coherent, and wrong in a way that changes how the situation is understood. How AI Systems Separate Content from Source AI systems do not evaluate information the way a reader traces a statement back to its origin. They ingest large volumes of distributed content, break that content into fragments, and recombine those fragments based on statistical relevance. In this process, repetition becomes a strong signal.…