OSINT Experts Warn AI Models Are Ingesting Poisoned Evidence

Open-source intelligence practitioners face an escalating crisis as AI-generated disinformation becomes indexed within searchable evidence databases, according to a warning issued by analyst JP Zanders. Zanders argues that synthetic media, including deepfakes and fabricated documents, is being systematically ingested by search engines, large language models, and archival systems used by journalists and law enforcement. This convergence transforms isolated fake content into persistent, retrievable data points that contaminate future investigations and model outputs. The concern extends beyond individual deceptive assets to the systemic pollution of the open web's information ecosystem. As AI systems increasingly rely on scraped web data for training and retrieval, the distinction between verified fact and manufactured narrative blurs within technical infrastructure. This development potentially undermines the foundational verification processes required for credible digital forensics and national security analysis.